Abstract
Care-providing parents of immigrants fill a labor market niche in the receiving economy. My research examines the intergenerational support for working women in immigrant families in the United States. Using panel data derived from the Current Population Survey (CPS), I find that having a coresiding parent increases the labor force participation probability of foreign-born women with children by about 7.4 percent, and that the effects differ by birth region and educational level. I use a difference-in-differences approach to reinforce the finding that coresiding parents significantly alleviate the short-term labor force participation decline of foreign-born females after their transition to motherhood.
- niches
- parents of immigrants
- caregiving grandparent
- intergenerational support
- female labor force participation
The prevalence of intergenerational support for working women has been documented in many countries and cultures. When immigrants arrive and work in the United States, this support is likely to follow. In my research, I have investigated this transnationalized phenomenon and quantified the important yet neglected role of immigrants’ parents in supporting their children in the U.S. labor market.
The main mechanism through which caregiving parents of immigrants increase female labor force participation is by increasing the availability of child care and reducing the cost of child care. As James Heckman (1974) found in his study of the labor force participation choices of mothers, living with a relative decreases the quality-adjusted child care price for white American families by 67 percent. In addition, grandparent-provided child care is usually trustworthy and considered the “next best thing” (compared to mother’s care) (Falk and Falk 2002; Wheelock and Jones 2002).
Since intergenerational support for working women is not a new phenomenon, it is useful to begin with the literature of intergenerational support in different countries and cultures and then look at how immigration has transnationalized it into the U.S. labor market.
The evidence on this topic from Europe is abundant and often focuses on regional differences. Many European studies are comparative analyses using panel data to control for unobserved family factors (Aassve, Arpino, and Goisis 2012; Albertini, Kohli, and Vogel 2007; García-Morán and Kuehn 2013; Hank and Buber 2008). In addition to confirming that grandparent-provided child care affects female labor force participation, they all agree that grandparenting practices vary among the countries under study. Family ties in southern Europe are known to be stronger than in the rest of Europe (Albuquerque and Passos 2010). Marco Albertini, Martin Kohli, and Claudia Vogel (2007) note that family ties in southern European countries are different from family ties in Continental or Nordic countries. In southern Europe, multigenerational families are more common and more resource exchange takes place between the generations. Studying ten European countries, Karsten Hank and Isabella Buber (2008) discover, using the Survey of Health, Ageing, and Retirement in Europe (SHARE), that the pattern may be more complex than a dichotomy of “strong versus weak” or “Scandinavian versus Mediterranean.” They find three distinctive regional groups in terms of grandparent caring: Danish, Dutch, French, and Swedish grandparents are the most likely to provide care but the least likely to provide regular care; grandparents in the Mediterranean countries are less likely to provide care, but the most likely to provide regular care; and Austrian, German, and Swiss grandparents are in the middle. The SHARE data also reveal that European grandparents vary with respect to whether they feel obligated to provide care to their young grandchildren: in the Netherlands and Denmark, fewer than 60 percent of grandparents agree that it is their duty to provide weekly care, but 95 percent of Greek grandparents see caring for their grandchildren as their duty.
Naohiro Ogawa and John F. Ermisch (1996) document the child-caring role of coresiding grandparents in Japan, an established industrialized country that suffers from low female labor force participation. Margaret Maurer-Fazio and her colleagues (2009) examine the topic in the context of rural-to-urban migration in China, using Chinese census data from 1982 to 2000. They conclude that coresidency with parents or in-laws increases the labor force participation rate of nonmigrant urban women by 4.6 percentage points. Maurer-Fazio and her colleagues hypothesize that because housing constraints have not been a major issue for most urban Chinese families since around 2000, elderly parents who coreside with their children are more likely to be frail and therefore add to the burden of the working female in the family. Thus, their finding of a positive impact of the presence of parents and in-laws on labor market outcomes is unexpected. Because in broader Asian culture women are expected to do the housework, I would expect to find a more significant supporting role for working mothers being filled by grandmothers—particularly the maternal grandmother, since she typically does the housework that is considered the mother’s responsibility.
Although the impact of intergenerational support on female labor force participation has been well studied, research has not examined such intergenerational support in the light of immigration. Care-providing parents of immigrants play a unique role in the receiving economy. My research unifies the intergenerational support practices of different countries and cultures in the U.S. labor market context. It also contributes to the immigration literature by providing possible explanations for the labor force participation rates of foreign-born females and the relationship between those rates and the level of intergenerational support they receive.
Destination countries and home countries may differ not only in their grandparenting practices but also in the impacts of those practices. First, the international immigration process, both permanent and temporary, is characterized by strong positive selection: the most willing and able grandparents come to the United States to offer care. Hence, I expect to find a larger positive effect of grandparental support on the immigrant mothers’ labor force participation compared to the non-immigrant coresiding mothers. Second, foreign-born grandparents who follow their children to the United States—that is, who are not first-generation or primary immigrants—are less independent financially and psychologically and more likely than their American counterparts and their counterparts in the home country to coreside with their children. Third, the help-receiving immigrants may have less competition for this parental support from siblings, owing to geographic separation from them. Finally, foreign-born grandparents’ age at migration, language restrictions, citizenship status, and low mobility (from an inability to drive) make them less likely than native grandparents to be in the paid labor force and, therefore, more likely to contribute to child care and other housework.
Since the 1990s, the age composition of immigrants to the United States has changed: the share of seniors (ages sixty-five and older) has increased from 9 to 12 percent. Senior immigrants also have a growing representation in legal permanent resident (LPR) admissions, in part because, as Jeanne Batalova (2012) points out, previous immigrants took advantage of U.S. immigration policy’s family preference and sponsored their parents to come to this country.
In examining the extent to which intergenerational support exists in U.S. immigrant families, such support should not only be put in the context of the immigrants’ ethnic groups in the home country but also analyzed in comparison with the host country. Part of that context in the United States is its cultural emphasis on independence and the nuclear family. In addition, the geographical distance between the host and home countries may severely limit grandparents’ ability to provide child care and household assistance (Fields, O’Connell, and Downs 2011). Eva García-Morán and Zoë Kuehn (2013) exclude the foreign-born from the samples in their study using German data because both “availability of child care by relatives” and “residence relative to parents” may be determined by very different factors. Studies of intergenerational support using U.S. data are sparse. Josefine Posadas and Marian Vidal-Fernandez (2013) quantify that grandparents’ provision of child care increases female labor force participation rate by 15 percent.
In my research, I compare the education levels of immigrant mothers who receive intergenerational help. Recent literature diverges in its assessment of which level of education brings a bigger motherhood penalty for women. Yanka Byker’s study (2016) shows that new mothers at the very high end of the education scale (master’s degree or higher) and those at the low-middle and low end (non-college-educated) experience steeper drops in labor force participation than those in the middle (bachelor’s degree). Paula England and her colleagues (2016) find that high-skilled, high-paid women experience the largest motherhood penalty, and Olena Nizalova, Tamara Sliusarenko, and Solomiya Shpak (2016), using data from Ukraine, find that low-educated women suffer the largest motherhood penalty. In looking at family networks and low-skilled immigration, Patricia Cortés and José Tessada (2011) find that low-skilled immigration increases paid labor market work hours for women at the top quartile of the pay scale because the economic return for highly educated females is larger in that market. However, evidence for the opposite impact can also be found: in her work using U.K. Time Use Survey data, Anne Grey (2005) concludes that there is a significant difference between mothers’ labor force participation with and without grandparental help, and that this difference is especially large for non-college-educated mothers.
IDENTIFYING THE NICHE
Susan Eckstein and Giovanni Peri (this issue) define an immigrant niche as “occupations in which a high percentage of workers are foreign-born.” If providing nonpaid, contingent child care is an “occupation,” then immigrants’ parents definitely occupy a niche: a higher percentage of foreign-born than native-born grandparents coreside with and provide care to their grandchildren. Compared to other labor market niches discussed in this issue, the niche occupied by immigrants’ care-providing parents is “flatter”: these grandparents are a relatively homogenous group, they are older, and their expected “work” time is short. Once the children are no longer dependents, and as these immigrants’ parents become older, they may receive care themselves and become consumers of welfare if they stay in the country.
The support of their own parents may free immigrants who are parents (most likely the mothers) from child care and household chores and allow them to be engaged in the paid labor force or to pursue the type of work that best matches their skills. This effect is similar to the complementarity effect observed between new low-skilled immigrants, on the one hand, and high-skilled natives and existing immigrants, on the other (Borjas 1999; Peri and Sparber 2009, 2011). The complementarity effect within families is even more direct and efficient because the inflow of foreign-born grandparents is totally demand-driven—foreign-born parents are always tied to an existing immigrant family, and there is a perfectly matched position for them. Research has documented that the presence of grandparents in immigrant households has a positive impact on the children’s language development and assimilation (Tran 2010), as well as on the maintenance of family and ethnic ties (Waters et al. 2011). In the extreme case where a new mother trusts only her own mother to look after the baby (Falk and Falk 2002), the market cannot provide alternative care and this niche will go unfilled if the grandmother is absent. Even if grandparents’ work could be replaced with day care services, one could argue that, given the shortage of day care services in most U.S. cities, the inflow of caregiving grandparents, rather than taking jobs from American day care workers, probably adds new day care “workers” to the market.
In discussing grandparent care preferences in their comparative study of European countries, Arnstein Aassve and his colleagues (2012) develop a typology of families in terms of the mother’s labor force participation and the grandmother’s care preferences: (1) modern families with highly motivated working women, weak family ties, and working grandmothers; (2) traditional families with a negative preference for working mothers, strong family ties, and a preference for family child care; and (3) mixed families with highly motivated working mothers mismatched with the available grandmothers. In U.S. immigrant families, the first family type (modern) may not be observed as often, since those grandmothers who prefer working or other retired activities other than caring for grandchildren are less likely to end up in the United States. Families with coresiding grandparents are likely to be either the second type (traditional) or the third (mixed), and the most important factor distinguishing these two types of families may be the mother’s education level.
This typological analysis points to two competing mechanisms that my research tests empirically: (1) in traditional immigrant families, the presence of a grandparent will not increase the mother’s labor force participation; and (2) in mixed immigrant families, although mothers and grandparents come from the same country and culture and may have similar preferences, being primary immigrants and more likely to be risk-seeking makes it more likely that these immigrant mothers are the modern type of mother, and therefore that intergenerational support will increase their labor force participation.
To explore regional differences, figure 1 illustrates the correlation of grandparent-provided care (proxied by a grandparent’s presence in the same household) with the labor force participation of immigrant mothers by region. If we consider the U.S. average as the original and divide the panel into four quadrants, we have three groups of regions with different preferences for the working and caretaking roles of mothers and grandparents. The upper-right quadrant, where both the percentage of households with a coresiding grandparent (GPinhh) and immigrant mothers’ labor force participation rate (FLPR) are higher than the U.S. average, encompasses those regions represented by the traditional type of grandparents and modern mothers (mixed families). East and Southeast Asia and Africa fall into this category. Regions in the lower-right quadrant have larger percentages of coresiding grandparents in the household and lower female labor force participation rates (traditional families). Central and South America, Southwest Asia, Southern Europe, and the Middle East fall into this category. In the lower-left quadrant—with rates lower than the U.S. average of coresiding grandparents and female labor force participation rates—we find the modern type of grandparents mixed with traditional mothers (Central and Eastern Europe, Western Europe, Northern Europe, and North America, not including Mexico). No regions are located in the upper-left quadrant. Note that this division applies only to immigrants in the United States from those regions, not to the actual populations of those regions.
Correlation of Labor Force Participation of Immigrant Mothers Ages Sixteen and Older with the Percentage of Households with a Coresiding Grandparent, by Region of Birth, 2014
DATA AND METHODS
To examine the relationship between grandparents’ presence and female immigrants’ labor force participation, the empirical analysis uses CPS data sets obtained through the Integrated Public Use Microdata Series (IPUMS). Managed by the Bureau of Labor Statistics (BLS), the CPS follows one family for four months and then, eight months later, follows up with the family for another four months. Thus, each family is interviewed for a total of eight months during a span of sixteen months. Taking advantage of the CPS’s longitudinal design, I construct a panel of linked households across interview periods using the unique household and person identifications produced by Julia Drew, Sarah Flood, and John Warren (2014) for IPUMS CPS data.
My research examines the effects of the non-first-generation foreign-born grandparents’ coresidence on the labor force participation of primary (first-generation) female immigrants by focusing on foreign-born female household heads and spouses, who are more likely to be primary immigrants.
The key explanatory variable, GPinhh (grandparent in the same household), is created by using information on household relations. (I recode relations into generational marks, then identify multigenerational families as GPinhh.) A grandparent’s presence in any given family can vary from period to period, as can the dependent variable, female labor force participation. Because CPS data do not contain information about whether a coresiding grandparent provides care for the children in the household, I make the plausible assumption that coresiding grandparents are providing such care and also helping with household chores to some extent. This assumption is supported by the positive selection inherent in the international migration process: the most willing and able grandparents are most likely to migrate. Although the presence of non-coresiding grandparents who live nearby and provide occasional or regular care has been documented to have a positive impact, the CPS does not provide this information, and therefore it is not possible for me to examine the impacts of non-coresiding grandparents.1 As discussed earlier, foreign-born grandparents who are nonprimary immigrants are more likely to live with their children; unfortunately, those who reside elsewhere are not included in this research.
Panel Regression
The first empirical approach I take is a panel regression using CPS data. For convenience, I use a linear probability model (LPM) for panel data with fixed effects for a binary dependent variable (labor force participation of immigrant mothers).
Since our key independent variable is also binary (having a parent in the same household), using an LPM is more reasonable than an ordinary least squares (OLS) model, and the “fitted probabilities are simply the average yi within each cell defined by the different values of x and no need to worry about probabilities less than zero or greater than one” (Wooldridge 2002, 454–57).
Let i denote an individual woman who is age eighteen or older, foreign-born, the mother of preschool-age children (under age six), and either a household head or the spouse of a household head, and let t denote the combination of year and month (the variable “mish”). Yi,t indicates the labor force participation outcome of women i at time t; GPinhhi,t indicates whether a grandparent is in the same household of women i at time t; Xi,t presents a vector of time-varying characteristics that include age, age squared, years of education, citizenship status (only for foeign-born: noncitizen or naturalized citizen), marriage status (dummy), birth region, and family income. Individual and time fixed effects are included in order to control the individual-invariant characteristics, such as location and personal preferences, as well as the time-invariant factors, such as the overall labor market environment.
I group birth countries into nine regions according to geographic adjacency as well as differences in European grandparenting practices, as suggested by the literature (Hank and Buber 2008): North America (excluding the United States), Central and South America, Northern Europe, Western Europe, Southern Europe, Central and Eastern Europe plus the Russian Empire, East and South Asia, India and Southwest Asia, the Middle East, and Africa.
Construction of the Panel Data and the Instrumental Variable
To better establish the causal relationship between a coresiding grandparent and a mother in the labor force, the previously used instrumental variable (IV) was the maternal grandmother being alive (Aassve, Arpino, and Goisis 2012; Posadas and Vidal-Fernandez 2013). Since this information is not provided by the CPS, I choose to use the source country’s retirement age as IV. I argue that earlier retirement directly increases the availability of grandparents and that retired grandparents are more likely to travel to the United States to coreside with their adult immigrant children and care for their grandchildren. At the same time, the retirement age in the parents’ source country is not likely to affect the labor force participation (dependent variable) of female household heads or spouses in the United States. I use retirement age data on sixty-seven countries, ranging from age fifty (China) to age sixty-seven (Iceland, Norway, and Greece), with some countries setting the retirement age in increments of 0.3 years (four months) and 0.4 years (three months).
In many cases, the grandmother came to the United States alone or, if she traveled with her husband, stayed for a longer period than her husband. Both child care experience and retirement age are factors in explaining why grandmothers more often provide child care, and for longer periods, than grandfathers. Because grandmothers outnumber grandfathers and are more likely to provide care when they coreside with children, when the retirement age for a source country differs for males and females—for example, in China women retire at age fifty-five while men retire at sixty—I choose to use the female retirement age.
Technically, testing the effect of having a grandparent in the household on labor force participation requires a binary panel data model with endogeneity. To keep the methods simple, I choose to implement panel regression with fixed effects using two-stage least squares (2SLS) with instruments for both binary outcomes (labor force participation) and continuous outcomes (hours worked last week).
Table 1 presents the descriptive statistics of the dependent and independent variables of the panel regressions. The variable “in labor force” for foreign-born mothers (47.9 percent) is substantially lower than that for natives (67.8 percent), with the standard deviation around 0.5 (50 percent) for both. The logged hours worked last week for those in the labor force are relatively equal (3.41 and 3.49). Foreign-born mothers are almost twice as likely to have a parent present in the same household as native mothers (5.4 percent versus 2.9 percent). Foreign-born mothers are also more likely than natives to be married (87.4 percent versus 77.9 percent). With an average of 14.0 years of education, native mothers are better-educated overall compared to foreign-born mothers (12.4 years), but the standard deviation for foreign-born mothers is larger, reflecting a more widespread distribution. High education (HE) and medium education (ME) dummies are defined as “college degree and above” and “some college education but no degree.” A third education category, low education (LE)—defined as “high school diploma and less than high school education”—is omitted by the regression owing to collinearity. Indeed, a much larger portion of foreign-born mothers belong to the low-education group (53 percent) compared to natives (29.6 percent).
The Impact of a Coresiding Grandparent on the Female Labor Force Participation Rate: Panel Regression Descriptive Statistics
The Difference-in-Differences Method: Following Women Before and After They Give Birth
Another empirical approach I use is a difference-in-differences model to identify the effect of a grandparent on new (immigrant) mothers. Handling “having a child” and “having a coresiding parent” as two independent treatments, this approach improves on the panel regression, where only mothers are examined, by allowing me to examine both women with children and those without children. Specifically, using 2006–2015 CPS data, I identify the month (interview period) in which a woman gave birth and mark all months after that as postpartum months. The other treatment—whether or not a parent resides in the same home—is the same variable as in the panel regression. The two treatments are independent, which means that they may or may not happen during the same period. Because the CPS follows individuals only over a span of sixteen months, this model focuses only on women with a child under two years old. (Assuming that a woman gives birth in the second period during which she and her family are interviewed by the CPS, the baby will be fifteen months old in the last period in which they are interviewed.) Most of the new births happen during the eight months when families are not interviewed. In other words, the period in which I see a change in the number of children is not likely to be the month in which the new baby was born. Hence, paid maternity leave is less of a concern for disrupting regression results. It would be ideal to know the exact months when women gave birth, but this information is not provided in the CPS.
The difference-in-differences model is constructed as a panel linear probability regression:
Yi_t indicates the labor force participation outcome—whether a woman is in the labor force (dummy variable) and hours worked per week at all jobs—of women i at time t. (Notice that this is different from Yi,t in a panel regression.) GPinhhi_t indicates whether a parent lives in the same household as a woman i at time t; Childi_t indicates whether a woman i is in a postpartum month at time t; GPinhh*Childi_t is the interaction term of having a coresiding parent and having given birth; Xi_t presents a vector of time-varying characteristics that include age, years of education, naturalized citizen (dummy), and marriage status (dummy). The coefficient, α3, is the difference-in-differences estimator.
Table 2 presents the descriptive statistics of the dependent and independent variables of the difference-in-differences regressions. The sample contains eighteen- to forty-five-year-old foreign-born females who are household heads or spouses of household heads. For a better before-and-after comparison, only individuals appearing eight times are included. A total of 18,705 women are included in the regression, and about 14.7 percent of them gave birth during the period followed in the data.
The Impact of a Coresiding Grandparent on the Labor Force Participation Rate of Foreign-Born Mothers: Difference-in-Differences Regression Descriptive Statistics
EMPIRICAL RESULTS
Regression Results
Table 3 presents the results for equation 1 (panel LPM regression with fixed effects) using CPS all-month data from 2006 to 2015. Except for column 4, all other models’ standard errors are clustered. Column 1 shows results for the labor force participation of native-born mothers who have at least one preschool-age child and are household heads or spouses of household heads. The key explanatory variable, “parent present in the same household,” has a significantly negative effect on labor force participation: with everything else controlled, having a coresiding parent in the household is estimated to reduce the probability of participating in the labor force by about 4 percent (0.042). This may reflect the American culture and also the negative selection of elderly coresiding grandparents: women may opt out of the labor force because they are more likely to care for coresiding parents than to get help from them. In addition, being married and having one additional preschooler are estimated to reduce the probability of participating in the labor force by 11 and 8 percent, respectively.
Female Immigrants’ Labor Force Participation Response, by Parents’ Presence
Column 2 shows results for the labor force participation of foreign-born mothers of at least one preschool-age child who are household heads or household heads’ spouses. Everything else controlled, having a coresiding parent significantly increases the probability of participating in the labor force for a foreign-born female with a preschooler by about 7 percent (0.074). This result is highly significant. The magnitude of this effect of intergenerational support on female labor force participation falls in between what Posadas and Vidal-Fernandez (2013) find (15 percent) using Southern European data and what Maurer-Fazio and her colleagues (2009) find (4.6 percent) using Chinese data. This result is plausible considering that the U.S. immigration population includes so many nationalities and cultural norms. Later in my analysis, regional interactions will shed light on this issue. Unlike Posadas and Vidal-Fernandez, who can clearly identify care-providing grandparents in their study, I have to proxy coresiding with care-providing, and this may discount the effect to some extent. That I find a larger effect than what Maurer-Fazio and her colleagues find may be due to the strong positive selection that international migration offers: compared to coresiding grandparents in native households, grandparents who are willing and able to travel internationally are more likely to provide care and to enable women to join the paid labor force.
In addition, column 2 shows that being a naturalized citizen (compared to being a noncitizen or legal permanent resident) increases the probability of participating in the labor force by 15 percent. For immigrant mothers, being married and having one additional preschooler are estimated to reduce the probability of participating in the labor force by 22 and 8 percent, respectively. This model also controls for the immigrant’s birth region; the results show that mothers from India, Southwest Asia, and the Middle East are less likely to be in the labor force, while those from Southern Europe and Africa are more likely to be in the labor force.
Column 3 uses the same variables as column 2, but here the key explanatory variable is instrumented. The coefficient of “parent” on the second stage becomes implausibly large, denoting a weak instrument. The fitness of the model is also poor. Table 4 shows the first-stage estimates of this 2SLS. The F-statistic of the first-stage estimates is 37.7, well above the threshold of IV qualification (10.0). “Parent” is statistically significant, but the coefficient is tiny (−0.002). Retirement age is a weak IV, possibly because variation (which is not large) occurs only by country, not by year. Also, working grandparents’ eligibility in many countries to take family leave to visit and take care of their U.S.-based grandchildren may also weaken the effectiveness of using retirement age as an IV in this analysis.
First-Stage Estimates of the Relationship Between the Probability of a Parent’s Presence and Retirement Age in a Fixed-Effects Model
Columns 4 and 5 of table 3 combine the regional dummy and each region’s interaction terms with “parent” to reveal the regional difference in grandparental support and its impact on mothers’ labor force participation in the United States without clustered standard errors (column 4) and with clustered standard errors (column 5). In the model without clustered standard errors, the regional interaction terms of “India and Southwest Asia” and “East and Southeast Asia” have significantly positive effects. Considering that preschoolers’ mothers from these two regions have a lower probability of participating in the labor force, having a parent in the household is estimated to increase the probability of the mother being in the labor force for families from these two regions. In other words, for immigrant women from these two regions, grandparental support has a fairly large positive effect on their labor force participation. However, with the standard errors clustered, all regional interactions’ significant effects disappear (column 5).
Finally, the regression on column 6 of table 3 includes college-educated (HE) and medium-educated (ME) dummies and allows interactions between these education indicators and a parent’s presence in the same household in order to examine whether parental support helps highly educated immigrant women more than their less-educated counterparts. The interaction term of college-educated and parent’s presence being highly significant indicates that the effects of grandparental support on labor force participation for college-educated immigrant mothers are larger than the effects on their less-educated counterparts. Everything else controlled, a college-educated immigrant mother living with a parent has a 16.8 percent higher probability of being in the labor force than an immigrant mother with a high school education or less who lives with a parent.2 This is also evidence that the labor force niche of immigrants’ parents has an economic rationale: it is more economically beneficial for a grandparent to provide child care and enable the higher-educated mother to work because highly educated women are paid more in the labor market. The interaction of being medium-educated with having a parent present in the household is not statistically significant; thus, the benefit from having a grandparent present is significant only when the mother is college-educated. This finding echoes the research by Cortés and Tessada (2011), who find that low-skilled immigration increases work hours for highly paid and high-skilled women.
To test the robustness I ran the same sets of fixed-effects panel OLS regressions on logged hours worked last week. The results (signs, significance levels, and relative magnitudes of the key explanatory variable) are similar. (Results are not shown here but are available upon request.)
Difference-in-Differences Regression Results
Table 5 shows the unweighted relations of the two treatments—having a newborn and having a coresiding parent. Foreign-born females from eighteen to forty-five years old who gave birth during the sixteen-month interview period are included. The labor force participation rate of the group with no parent present in the same household drops about five percentage points postpartum, compared to almost no change for the group with a parent present.
Labor Force Participation Rate of Foreign-Born Women Ages Eighteen to Forty-Five Before and After Giving Birth, by Parent Presence in the Same Household
Table 6 presents the results of the difference-in-differences method implemented by a regression with interaction terms of the two treatments. Column 1 shows that the difference-in-differences estimator is statistically significant at the 5 percent level. The coefficients of the first three explanatory variables show that for a foreign-born woman who does not have a coresiding parent, the probability of being in the labor force declines 6.7 percent after she has a child, compared to a decline of 3.4 percent for those who have a coresiding parent. In other words, having a coresiding parent significantly alleviates the decline in labor force participation of new immigrant mothers.
Female Immigrants’ Labor Force Participation: Difference-in-Differences Regression Results
Column 2 presents the results of the same model run on the native-born counterparts of these foreign-born mothers. Both having a newborn and having a coresiding parent decrease the probability of being in the labor force for a native mother. The difference-in-differences estimator is positive (meaning having a coresiding parent may alleviate the postpartum labor force participation decline), but it is not statistically significant.
When I run the same regressions using logged hours worked as the dependent variable, the difference-in-differences estimator is not statistically significant. The effect on hours worked may not be as significant as on labor force participation because, with the CPS data following women for only a relatively short period (sixteen months), there might not be much difference in work hours for women who have already shortened their work hours prior to giving birth. On the other hand, pregnant women who intend to stop working are likely to stay in their job (although cutting down their hours) until the baby’s arrival and then choose not return to work. Therefore, using CPS data and this research design, labor force participation is a better outcome to examine rather than hours worked.
NON-IMMIGRANT PARENTS AND THE CASE OF CHINA
More recently arrived parents of immigrants are more likely to be on non-immigrant visas (usually the B-2 type), and they are not well represented in CPS or census data. As discussed earlier, the non-immigrant parents of immigrants can obtain LPR status and eventually become naturalized citizens through family sponsorship from their immigrant children. Not much attention has been paid, however, to the recent rise in the number of older temporary visitors. We can better understand the current magnitude and the future trend of this caregiving population by examining the non-immigration visa data.
Figure 2 shows the recent fifteen-year trends in B-2 admissions for those age fifty and older from selected countries. B-2 visas are issued for tourists and for those visiting friends and family in the United States temporarily. Given that we do not have detailed information about the purpose of these visits, we have to assume that some of these visitors are care-providing grandparents. Numbers for these grandparents are mixed in with those for tourists and temporary visitors in this age group. While the care-providing visitors typically stay longer than others (usually up to six months, the maximum length of stay allowed by this visa), we have no specific information about their length of stay that would give us a better estimate of their numbers.
B-2 Admissions for Tourists and Visitors Age Fifty and Older from Selected Countries (Not Including Mexico), 2000–2014
In figure 2, data for Mexico, which has experienced the most drastic increase in B-2 admissions,3 has been graphed separately so as not to dwarf the changes in B-2 admissions from the other countries. China, Brazil, and India are the other three countries that have recently sent a soaring number of older B-2 visitors to the United States—a trend that echoes the observation earlier in this article that the cultures and practices of Latin America and East and Southeast Asian countries are among the most noteworthy in providing grandparental support. The dramatic increases in older B-2 visitors may also be explained by the rise of the middle class in these countries, as well as the existing immigrant population from these countries in the United States. It is also interesting to compare China with Taiwan and South Korea, two other countries that have similar Confucian cultures and that emphasize intergenerational support. The economies and immigration inflows to the United States of these two countries took off prior to the more recent period of Chinese economic development and increasing immigration inflows, their fertility rates have been low, and their “grandparent” inflow has been stable. (We even see a decline in the number of older B-2 visitors from South Korea.) Examining the older B-2 visitor trends helps us understand how the caregiving grandparents’ niche is perpetuated by family ties to earlier immigrants as well as by the economic development of the major immigrant-sending countries.
I examine China not only because of its grandparental culture but also because, in the current Chinese economic and political context, more parents of current Chinese immigrants are expected to arrive. Chinese elderly are highly involved with their grandchildren. Studies have found that this tradition is maintained in Chinese immigrant families in the United States and that grandparents in Chinese immigrant families do more than take care of children and help with household chores: they also educate children about Chinese language and culture (Xie and Xia 2011), provide role models, and maintain the family’s historical continuity (Falk and Falk 2002, 134–36).
Many Chinese immigrants who arrived in the last couple of decades were at the prime age for getting an education or taking a high-skill job. Immigrants in this wave were mostly born after the implementation of China’s one-child policy and are likely to be the only child in their family; as such, they are accustomed to their parents’ care. As these highly educated immigrants settle in the United States and have their own children, their parents follow as caregivers. In light of the financial success of the Chinese middle class, it is likely that the number of migrating Chinese grandparents will continue to increase in the near future.
To supplement the quantitative analysis of non-immigrant, care-providing Chinese grandparents, I conducted semistructured interviews with fifteen Chinese immigrant families that include grandparents who are on B-2 visas. Although the limited sample does not permit generalization, this qualitative part of the research allows me to describe the dynamics of this special population in greater detail.
Yang came to the United States for an advanced degree in 2004 and currently holds a full-time job on the East Coast. Both Yang and her highly educated husband, also born in China, are the only children in their families. Even when Yang was in school, her mother would come to the United States for a couple of months to provide care and cook for her during critical exam times. When Yang had her first daughter, their family had the typical “4-2-1” family structure of Chinese households after the implementation of the one-child policy (which stayed in place for over a generation): four grandparents, two parents, and one child. Recently Yang and her husband had a second daughter. Ever since Yang’s older child was born, her parents (mostly her mother) and her in-laws have been taking turns staying with the family, each for around six months, and this pattern has continued with the addition to the family. Besides taking care of the kids, her parents and her in-laws also cook and help around the house when they have a chance. Yang’s older daughter, now three, speaks mostly Chinese so far, since Chinese is the only language spoken at home. Yang considers herself very lucky to have both sets of parents providing care: “It saves us so much money. If not for their help, I don’t think it would make sense for me to go to work while sending both kids to day care. I told them they do not need to take care of the elder one after she turns two, and was thinking about sending her to day care. But they insist on keeping her at home for longer.” Yang and her husband are legal permanent residents and in the process of being naturalized. They plan to sponsor their parents for LPR status as soon as they become eligible.
Wang and her two-year-old son followed her husband, who took a faculty position at a Midwestern U.S. university in 2011. Wang has been working on a PhD degree at the same university. Their son was born in Singapore in 2009, when Wang’s husband held a postdoc position there, and Wang’s mother went to Singapore to take care of her and the baby. Later, they moved back to a city not far from Wang’s hometown in China, and her mother often visited them there to help out. Since they moved to the United States, her mother has visited three times, each visit lasting three to six months. Wang’s daughter was born in 2015. Because her mother was experiencing health issues at that time, her in-laws came from China to stay with them right before the baby arrived to provide care. During that period, her father-in-law’s own father in China became sick and he had to return. Her mother-in-law stayed for the whole six months allowed by her B-2 visa. Wang has a younger brother who has stayed in Canada after going there for his master’s degree. He and his wife are expecting a child, and Wang’s mother plans to help them when the new grandchild arrives. She will stay for only two weeks, however, since her daughter-in-law’s parents will also be there and plan to be the primary caregivers.
In Wang’s case (and her brother’s), we see the strong preference for the maternal grandmother as the first choice of caregiver. The first month after childbirth, known as yuezi (meaning “month”), is crucial for both mother and baby in Chinese culture and practice. Because new mothers are expected to lie down whenever possible, they need help taking care of the baby as well as themselves. Traditionally, most women married into the husband’s family and the paternal grandmother was the main caregiver during the yuezi period. Now that more and more young couples live apart from their parents, however, the first choice for this role is the maternal grandparents—especially the maternal grandmother, because she knows best the new mother’s eating and living preferences.
Zhang came to the United States for an advanced degree in 2009. Her husband is a third-generation Chinese immigrant born and raised in a Latin American country. They recently moved from the South to the West Coast while Zhang was pregnant, and her husband took a job there. After their baby arrived, Zhang’s parents came to stay with them to take care of the baby, and when the baby was three months, Zhang happily went back to work in a new job. Without her parents’ support, according to Zhang, it would have been a lot more difficult for her to return to the labor force. “I know that they will do a good job, a better job than I perhaps,” said Zhang. She also mentioned that her in-laws, who are second-generation Chinese immigrants living in Latin America, are not likely to offer such child care, partly because they are still working in their country, but also because, being more Westernized, they are not used to the idea of offering to care for grandchildren, as Zhang’s parents are. Zhang has applied for an extended stay for her parents of another six months so that they can remain until her baby is close to a year old and more ready for day care.
In the absence of grandparents willing and able to provide care, my interviewees’ backup plans included care by the mother herself, care by the other set of grandparents, day care, and nanny service. No one mentioned “satellite babies” as an option: in an earlier prevalent practice, Chinese immigrants sent infants back to China to be raised by members of their extended families (Bohr and Tse 2009; Skeldon 1997; Waters 2001).
All child-caring grandparents I interviewed had close social network ties in China. Some had other grandchildren who also needed care, and some were caregivers for their own parents in China. Their double caregiving role and close social ties back in China required that they travel frequently. At the same time, their limited free time while providing care, cultural differences, language barriers, and the low walkability of American cities made Chinese grandparents more socially dependent during their time in the United States than they were back in China.
The one-child policy may have reduced the competition for attention from other grandchildren, but when the only child lives overseas, the China-based parents face a difficult choice about where to settle long-term. I gleaned from my interviews that the main consideration in determining the long-term plan is whether the immigrant is the only child. If they have other children in China, the parents tend to say that they will eventually go back to China. Those with an only child who lives in the United States, especially widowed parents, seem to be indecisive and concerned about the future. Huang, the only child of her family, came to the United States for her PhD, is now married to a U.S. citizen, and has a two-year-old son. She told me: “I grew increasingly worried about my mother after my father passed away four years ago. I feel obliged to call her every day to make sure she is doing well when she is in China. My mother is happier here to see us every day, but I can tell she still misses her sisters and friends at home. In another ten years, her friends and sisters may all leave our hometown and stay near their own children in other cities. It’s almost definite that she will need to stay with us in the future, although she says she is undecided.”
Other concerns about staying in the United States include the language barrier, lifestyle differences, lack of health insurance, and social dependency. Song, a sixty-two-year-old caregiving grandmother in the United States, told me that once when she had to be hospitalized, the doctor gave a long explanation and instructions to her through her daughter, but her daughter translated it into only a few sentences in Chinese. “I feel very frustrated, not being able to understand anything,” Song said. Liang, a caregiving grandfather I interviewed, said, “Even in China nowadays, young people do not want to live with their elderlies because it is not convenient that way. We do not want to live with them [his children] either. We are freer by ourselves. Not to mention in the United States we do many things differently.” Although most of the China-based grandparents I interviewed were middle-class, they spoke no English or very limited English. Most of them could not afford housing in a major U.S. city and also found it socially hard to maintain a household by themselves in the United States. In China, they might have been leading a happy retired life, just without their beloved grandchildren. The advantages of staying in the United States mentioned by these grandparents included being close to children and grandchildren, the nice weather and environment, better economic and educational opportunities for children and grandchildren, and readily available Chinese produce.
CONCLUSIONS AND DISCUSSION
Overall, I find that immigrants’ parents fill an important yet often overlooked niche in the U.S. labor market—providing care for their grandchildren and freeing up female immigrants to participate in the paid labor market. Although grandparental support is not a new phenomenon and has been analyzed in the literatures of many regions, it has not been examined and documented in the context of immigration before; nor has it received much attention from immigration researchers and policymakers. Care-providing parents of immigrants work for free, and their economic contribution is largely hidden. Foreign-born, care-providing grandparents are unpaid contingent workers in the U.S. labor market who fill a niche that otherwise might not be filled (in the case of stay-at-home mothers).
My empirical analysis using panel regression finds that having a coresiding grandparent increases the labor force participation probability of immigrant new mothers by about 7.4 percent and that the effects differ by birth region or country and educational level. Further, my difference-in-differences analysis shows that coresiding parents significantly alleviate the labor force participation decline of new immigrant mothers.
A case study of China-based care-providing parents illustrates that the role of immigrants’ parents goes beyond lifting female immigrants into the paid labor market. Over the long term, their presence and caregiving may improve the health of their immigrant children and grandchildren as well as the grandchildren’s school performance. The time and help that immigrants’ parents devote to the third generation also seem to increase their own mental health.
Because of limited data availability, this study assumes that coresiding grandparents provide care, since the CPS does not provide this information. Therefore, I have examined the impact of coresiding grandparents only, not of those who live nearby and provide care. Also, the short longitudinal coverage of the CPS confines the difference-in-differences analysis to children between ages zero and fifteen months. This is reasonable for examining the impact of grandparents visiting and providing temporary care, but it is not ideal for looking at those who are permanent legal residents—that is, the parents of children who have become naturalized citizens. As for the dependent variable, I use women’s labor force participation as a measurement of the impact of intergenerational support, on the assumption that coresiding grandparents provide child care to an extent that affects the mother’s decision to be employed in paid work. Though providing child care might not necessarily lift mothers to the paid labor force, help from coresiding grandparents could influence a number of other factors. Future research should examine other labor market outcomes, such as the family wage gap, gender inequality, and related health outcomes such as women’s fertility rate and postpartum health.
Finally, both the availability of immigrants’ parents and their future in this country are largely shaped by U.S. immigration policy. Many primary immigrants who are in their child-bearing years (the time when they need help the most) are work-visa holders, LPRs, or working on becoming naturalized, which is a lengthy process. The caregiving parents of this population arrive on B-2 visas and are restricted by the six-months-per-year rule (and three months for those from Visa Waiver Program [VWP] member countries, who are traveling without a visa). These parents’ path toward naturalization is determined by U.S. naturalization policy, which differs by country of birth and sponsor’s citizenship status. In addition to immigration status, foreign-born grandparents providing child care face many other challenges that call for policy responses, including health insurance coverage, language barriers, and adaptation to U.S. life. Even those grandparents who have become LPRs face the tough choice as they grow older and overseas travel becomes increasingly inconvenient between living with their children and grandchildren in the United States, with all the language and cultural barriers, and remaining in their home country and not being able to see their extended family in the United States—perhaps their only remaining family.
FOOTNOTES
↵1. On the positive impacts of non-co-residing grandparents, see two studies that use SHARE data: Hank and Buber 2008 and García-Morán and Kuehn 2013.
↵2. College (highly educated) dummy’s coefficient (0.144) plus interaction term’s coefficient (0.024).
↵3. This might in part be due by a change in counting methods by the Department of Homeland Security (DHS). Until 2005, typically only the initial land admission of an I-94 non-immigrant was recorded by the DHS. That year, the DHS began an effort to record all land admissions. For more information, see Department of Homeland Security, “The Impact of Counting Changes on Nonimmigrant Admissions: An Update,” 2012, available at: https://www.dhs.gov/sites/default/files/publications/Impact%20of%20Counting%20Changes%20on%20NI%20Admissions%20August%202012.pdf (accessed June 2, 2017).
- © 2018 Russell Sage Foundation. Hu, Xiaochu. 2018. “Filling the Niche: The Role of the Parents of Immigrants in the United States.” RSF: The Russell Sage Foundation Journal of the Social Sciences 4(1): 96–114. DOI: 10.7758/RSF.2018.4.1.06. Direct correspondence to: Xiaochu Hu at hxch.peking{at}gmail.com, University of the District of Columbia, Building 44, Room 119, 4200 Connecticut Avenue NW, Washington, D.C. 20008.
Open Access Policy: RSF: The Russell Sage Foundation Journal of the Social Sciences is an open access journal. This article is published under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.