Abstract

Recent decades have seen increases in the variability of family income, tepid income growth rates for all but the richest families, and widening income inequality. These trends are concerning for child well-being, given the importance of income to parental investments and parenting practices. Growing evidence suggests that a high level of change is disruptive to family processes and that chronic stress affects physiology as well as psychology. This study used the Panel Study of Income Dynamics Child Development Supplement to estimate associations between three dimensions of childhood income dynamics—level, variability, and trend—and child achievement and behavior. After income level was controlled for, income variability during childhood was not associated with child achievement or behavior, but an increasing five-year trend in income-to-needs was modestly beneficial to behavior measures. Subgroup analysis suggests some adverse effects of income variability and trend on reading and behavior for non-White children but no clear patterns by child's age or family income or wealth levels.

Introduction

The dynamic nature of family economic circumstances is a growing focus in multiple social science fields. In particular, recent studies documented high levels of variability in family income both between and within years, particularly for lower socioeconomic status (SES) and non-White families (Dynan et al. 2012; Hardy and Ziliak 2014; Western et al. 2016). This trend is consistent with—and likely driven by—growing instability in employment, family structure, and public assistance (Cavanagh and Fomby 2019; Ha et al. 2020; Hardy 2017; Kalleberg 2010). The disruptions to family life and the economy associated with the COVID-19 pandemic are poised to exacerbate these trends.

Although hundreds of studies across disciplines considered the influence of family income level on child well-being (for a review, see Gennetian et al. 2010), we know less about the effects of income dynamics on child development because so few data sources offer longitudinal measures at regular intervals of both family income and developmental outcomes. A handful of studies suggested that greater income variability has small negative effects on school outcomes, both in adolescence and in adulthood (Gennetian et al. 2018b; Gennetian et al. 2015; Hardy 2014). Important, unanswered questions include whether income variability and trend have effects on the development of younger children or in other domains, and whether those effects are moderated by race or SES.

Using the Panel Study of Income Dynamics (PSID) core and Child Development Supplement I (CDS) data, I estimated relations between income dynamics and child achievement and behavior. The PSID offers high-quality income and child outcome data, a rich set of covariates, and important potential moderators, including race and family wealth. Based on prior studies and insights from developmental psychology and neuroscience about the consequences of repeated change, I ask the following questions about income dynamics and child development:

  1. Are income variability and trend associated with child achievement and behavior after income level is controlled for?

  2. Are there threshold effects of high income variability, or negative or no income trend, on child achievement and behavior after income level is controlled for?

  3. Do the associations between income variability or trend and child achievement and behavior vary by child's age or race?

  4. Do the associations between income variability or trend and child achievement and behavior vary by family SES, as measured by income level or wealth?

This study contributes to an understanding of the dynamic nature of family life and particularly whether economic instability is a unique form of disadvantage. It distinguishes the salience of income level, variability, and trend for child development both conceptually and empirically. The analysis provides estimates of the effects of income variability and the trend in income-to-needs on developmental domains and age groups not previously studied. The results are essential for population science because income variability and other forms of instability have potential implications for processes of social stratification and mobility (Hacker 2008; Western et al. 2012) and for child well-being (Sandstrom and Huerta 2013). Importantly, U.S. income support programs focus largely on raising income level or subsidizing basic needs and may not be well designed to promote stability of economic circumstances during childhood.

Background

Income dynamics during the life course is a function of three dimensions: level, variability, and trend (Duncan 1988). In any period, income level is what economists refer to as permanent income, which captures the average resources available to the family across the entire period (Friedman 1957). Income variability or volatility (used synonymously here) is the change in income around the average (Dynan et al. 2012; Gottschalk and Moffitt 2009; Hardy 2017). Income trend is the direction and slope of income change from the beginning to the end of the period. Distinguishing income trend from income variability during childhood is critical because mobility itself is a form of variability, albeit one that may be beneficial to children.

Trends in Childhood Income Dynamics

Recent trends in childhood income dynamics reflect the population-level increases in income inequality. Childhood income levels have diverged over time by parents' education, with nearly all the growth since the 1970s among highly educated, high-income parents (Hill 2018; Lichter and Eggebeen 1993; Western et al. 2016). Earnings and income variability increased in that same period, particularly for lower-SES and non-White families (Dahl et al. 2011; Gottschalk and Moffitt 2009; Hardy 2012; Hardy and Ziliak 2014). These trends were driven by increasing divergence in employment and family structure stability by SES (Kalleberg 2010; McLanahan 2004; Western et al. 2016). Independent of economic cycles, both between- and within-job instability have increased for workers in low-paying jobs as employers shifted away from standard long-term employment arrangements toward a variety of contingent arrangements (Kalleberg 2010). Simultaneously, family structure changes during childhood—including divorce, repartnering, relationship churning, and coresidence with other family members—became normative, particularly for children with less-educated parents (Perkins 2017; Turney and Halpern-Meekin 2017). As employment and family instability increased, income support programs became less able to smooth family income (Ben-Ishai 2015; Hardy 2017).

Less is known about how income trends during childhood have changed over time. Hill (2008) found that the positive growth rate of family income during childhood declined from 3% or 4% in the 1970s and 1980s to 1% in the double-recession 2000s. On average, family and household incomes increase with child's age, primarily because parental earnings increase with age and work experience but also because adolescents can contribute work earnings to households (Duncan 1988). This age-normed expectation is supported by evidence that the risk of being poor declines for adults as they move from their 20s into middle age (and then increases again in later life; Sandoval 2009). In addition, life events such as parental death, divorce, incarceration, or a health crisis can drastically alter income trends during childhood.

Income Dynamics and Child Development

Higher income during childhood is beneficial to children in every domain of health and development (for reviews, see Bradley and Corwyn 2002; Case et al. 2002; Gennetian et al. 2010). Family income affects child development indirectly through parental investments and parenting practices. Parents can use higher income to purchase more and better food, housing, childcare, and schooling, all of which create the contexts for positive child health and development in multiple domains (Becker 1981; Bradley and Corwyn 2002). Also, more income leads to less stress, which is associated with warmer and more responsive parenting (the Family Economic Stress Model; e.g., Conger et al. 1992; McLoyd 1990; Mistry et al. 2009), a particularly salient pathway for effects on child social and emotional development (Guo and Harris 2000; Yeung et al. 2002). There is evidence of both linear and nonlinear effects of income on child achievement: an additional dollar is associated with improved outcomes, although so too is living above the poverty line, and an additional dollar has stronger effects if income level is low (Dearing et al. 2001; Duncan et al. 2012; Gennetian et al. 2010). Finally, income is also a mechanism by which other aspects of family life, such as family structure and parental employment, can affect children. In this study, however, I focus on income as a developmental context rather than as the mechanism for other contexts.

Conventionally, studies on income and child development “average out” variability in multiple income measures to better capture the resources available to families across time. Average income has been shown to relate more strongly than does any year's income level to family and child outcomes (e.g., Blau 1999; Dahl and Lochner 2012). Also, the permanent income hypothesis from economics posits that families can smooth income and consumption with savings when income varies (Friedman 1957). Importantly, this hypothesis is unlikely to hold for many low- and moderate-income families, or for African American and Hispanic families, who have less access to credit and far lower levels of wealth than White families (Dynarski and Gruber 1997; Hamilton and Darity 2017; McKernan et al. 2013).

Figure 1 is a heuristic that illustrates the need to move beyond either average income or measures of discrete changes in income to accurately capture and understand the importance of family income dynamics during childhood. Each of the five hypothetical patterns in the figure averages, across the four years, to the same yearly income of $22,000. Implicit in most prior research on income during childhood is the assumption that these patterns are equivalent vis-à-vis their effects on children.

Yet, the idea that variability in context could affect family processes and child development is centrally rooted in development theory. The bio-ecological model of development emphasizes the importance of consistent and predictable proximal processes—a child's interactions with people and things—in supporting healthy developmental growth (Bronfenbrenner 1995; Bronfenbrenner and Morris 2007). A child's exposure to proximal processes is defined by Bronfenbrenner and Morris (2007) along the dimensions of duration, intensity, and interruption. Direct examination of dynamic contexts of development has come in the study of environmental chaos, defined as contexts that are overstimulating because of disorder (e.g., crowding, noise, clutter) or instability (e.g., changes in family structure or parental employment; Evans and Wachs 2010). Both disorder and instability early in life are associated with child physical and mental health issues (Coley et al. 2015), but one study suggested that disorder—not instability—mediates the relationship between income and child achievement (Garrett-Peters et al. 2016). Similar to models of income level effects on child development, studies of chaos have also found that parenting is a key mediator, with reductions in responsivity and warmth and increases in harsh parenting being especially implicated in negative outcomes (Coldwell et al. 2006; Dumas et al. 2005; Vernon–Feagans et al. 2016). Recent advances in social neuroscience have also highlighted the potential for chronic or “toxic” stress to alter the body's stress response system (Ganzel and Morris 2011; Shonkoff et al. 2012). The human body is well designed to cope with intermittent or occasional stress but can be dysregulated by chronic or intense stress (Ganzel and Morris 2011). Importantly, economic instability may challenge both the child's stress response system and parenting quality, which is an important predictor of children's positive stress reactions.

Overall, if income variability reflects or creates too much change in children's lives—and particularly if it causes chronic stress and related reductions in warmth and contingency in parent-child interactions, or instability in other key developmental contexts, such as neighborhoods and schools—it could have long-lasting effects on stress response systems, emotional health, and social relationships (Danese and McEwen 2012; Evans et al. 2009; Miller et al. 2011). If, however, that variability is part of an upward trend in income leading to absolute improvements in material well-being and reductions in stress, or if the changes are associated with improvements in parental mental health or family functioning, income variability itself may not be detrimental (Hill et al. 2017). Although not the focus of this study, the nature of precipitating events may moderate the effects of income variability and trend on children. A parent's death or departure leading to the loss of a family's stable income source is likely to be experienced differently than income variability created by job changes or the loss of benefits (Hill et al. 2013; Sandstrom and Huerta 2013). There is limited evidence, for example, that poverty may be more detrimental to family processes and child well-being if it is associated with certain events or conditions, such as parental disability (Ratcliffe 2015).

Before these more nuanced relationships can be understood, however, research needs to account fully for the distinct effects of income variability and trends on child development, similar to the decades of studies documenting the overall effects of poverty and income level. A handful of studies have suggested that income variability may have adverse effects on parents and children. In a study of the pathways of income effects on child development, Yeung et al. (2002) found that large year-to-year income changes (30% or higher) were associated with an increase in maternal depression, decreased child test scores in math, and increased child behavior problems. Three recent studies found negative associations between continuous measures of income variability and both adolescent and adult school outcomes (Gennetian et al. 2018b; Gennetian et al. 2015; Hardy 2014). Using the Survey of Income and Program Participation 2004 panel, Gennetian et al. (2015) found that a greater number of intrayear income shocks over 32 months is associated with lower odds of a high level of engagement in school among adolescents. Using the Panel Study of Income Dynamics (PSID), Hardy (2014) found a small negative association between income variability during childhood and adult educational attainment. Notably, the associations between income variability and child achievement and behavior were consistently smaller than the associations between income level (permanent income) and those same outcomes.

Prior research suggests three potential moderators of the effects of income dynamics on child achievement and behavior: (1) child's age, (2) child's race, and (3) family income or wealth. First, although the few studies of income variability and child development have focused on adolescents (Gennetian et al. 2018b; Gennetian et al. 2015), developmental theory and prior evidence suggest that family economic circumstances matter most early in life because of age-specific developmental tasks, the high salience of the family context, and the cumulative nature of skill formation (Duncan et al. 2012; Heckman 2006; Shonkoff et al. 2012).

Second, income dynamics are not evenly distributed across racial groups, and the effects on children would not be expected to be evenly distributed, either. Income variability has increased across all racial/ethnic groups, but African American families face higher variability and less income growth than non-Hispanic White and Hispanic families (Gennetian et al. 2018a; Hardy 2017; Hill 2018). In addition, Black and Hispanic families face additive or intersecting disadvantages in the racial and economic hierarchies, which can alter the benefits that income confers, how income shapes the family and other developmental contexts, and the culturally adaptive practices used to support child development (García Coll et al. 1996; Henry et al. 2019). The few studies of income variability and adolescent school outcomes found larger adverse effects on adolescents of color (Gennetian et al. 2018b; Gennetian et al. 2015; Hardy 2014).

Finally, income or wealth levels could moderate the effects of income variability or trend but in ambiguous directions. Economic theory suggests that families with higher income or greater wealth should be able to withstand greater income variability without it affecting consumption or causing substantial stress (Friedman 1957). Indeed, income changes have stronger effects on child development when family income is low (Dearing et al. 2006; Duncan et al. 2012). However, recent studies found that disruptions to the family system—from paternal incarceration or maternal job loss, for example—matter the most to the children who are least likely to experience them (Brand and Thomas 2014; Turney 2015). If this were to hold for income variability, variability could be particularly impactful to children with relatively stable and sufficient resources.

Because the study of income dynamics and child development is nascent, I sought to document the basic associations between three components of income dynamics—income level, variability, and trends—and child achievement and behavior. Important, unanswered questions that guide this study include whether income variability and trends have effects on the development of younger children or in domains not previously studied and whether those effects are moderated by child's age, child's race, or family SES. Based on prior studies and theoretical insights about the consequences of repeated change, I hypothesized that with income level controlled for, income variability would be negatively associated with child developmental measures, particularly behavior, and income trend would be positively associated with those measures. In addition, I expected that these effects might be larger for younger, non-White, and lower-SES children.

Methods

Data and Sample

I estimated the associations between income dynamics and child achievement and behavior using the PSID individual, family, and Child Development Supplement (CDS) files. The PSID is the preeminent source of nationally representative survey data on intergenerational income and well-being in the United States (Duncan et al. 2018). Starting in 1997, the CDS followed a sample of children aged 0–12, with data collection occurring in 1997, 2002, and 2007 (Institute for Social Research n.d.). Table 1 shows the temporality of the measures used in this analysis, as well as sample sizes and child age ranges. The PSID-CDS sample with outcome measures was aged 3–14 at Wave 1, 6–17 at Wave 2, and 10–17 at Wave 3. The percentage of cases missing data varies from 0% to 10% in the variables used in the analysis. The total sample size is 7,042, with 3,485 unique children observed in at least one wave.1 Although this sample size was sufficiently large to estimate precise associations in the full sample, it did constrain subgroup analyses.

Analytic Approach

To identify associations between income dynamics and child achievement and behavior, I estimated the following ordinary least squares (OLS) model:
Yiw=α+βI1Piw5+β2Viw5+β3Tiw5+XiwδX+εi,

where Yiw was an achievement or behavioral outcome for child i at the time of the CDS survey wave (w). Piw –5, Viw –5, and Tiw –5 were child i's family income level, variability, and trend, respectively, all measured across the five years leading up to the CDS wave (w – 5). Because income level has both linear and threshold associations with child outcomes, I tested continuous and dichotomous measures of income variability and trend. Xiw is a vector of child and family-level control variables measured at the time of the CDS wave. εi captures unobserved heterogeneity in the model. All models adjusted standard errors for nonindependence between child-level observations. In addition, all descriptive statistics were weighted using the child-level weight from the CDS.

I examined the associations between dynamic income measures and child achievement and behavior for subgroups based on child's age and race. The relatively small sample size of the PSID-CDS constrained these analyses. To maintain reasonable subgroup sizes, I compared adolescents (13 years or older) with nonadolescents (3–12 years) and White children with non-White children. I show the descriptive results of separate subgroup regressions, but I also report the results of models using interactions to test the statistical significance of the difference between subgroups. Finally, I tested the significance of interactions between income level and variability and between wealth level and variability.

The estimates from this study should not be interpreted as causal relationships between income dynamics and child development. The estimates are unbiased only if all correlated factors associated with family economic context and child development were observed and controlled in the models. One common approach to controlling for unobserved time-invariant factors—fixed-effects estimation—was not possible here because capturing income dynamics requires a reasonably long period. For instance, income variability and trend were measured over five years, and I did not observe children long enough to take advantage of intrachild variation in those measures.

Measures

Table 2 provides weighted, cross-wave descriptive statistics on all variables used in this analysis, including five-year family income dynamics, child outcome measures, covariates, and moderators. Each of those variables is described in this section.

The PSID collects detailed information on all income sources—including earnings, transfer income, and other sources—for all adults in the family unit. The PSID-generated total income measure combines four primary categories of income: (1) head's and wife's taxable income (earnings plus other income from assets), (2) head's and wife's transfer income, (3) other persons' taxable income, and (4) other persons' transfer income.2 The PSID was conducted yearly until 1997 and biennially in subsequent years. The income measures capture the five years before the child outcome measure, using five observations for Wave 1 and three observations for Waves 2 and 3 (see Table 1). I top-coded total family income at 3 standard deviations from the mean of the overall PSID sample. All dollar values were inflated to 2011 dollars using the Consumer Price Index. For each child in the PSID, I created the following measures of family income dynamics.

Income Level

I calculated average family income across years for each child. I also used a binary measure of income level, indicating whether the family income was below or at or above the median income in the sample ($54,580).

Income Variability

Income variability was measured as the standard deviation of yearly arc percentage change (APC), a measure that has the advantages of being calculable with a zero value in one period and of producing symmetric values for positive and negative changes of the same size. APC was calculated in each year of the survey using the following equation:
APC=100.(YtYt1)/(Yt+Yt12),
(2)

where Y is family income; and t indexes years of data. APC is bounded by −200 and 200, and changes to and from 0 are measured as −200 or 200 APC. I also tested the sensitivity of the results to an alternative measure of income variability—the transitory variance of income—measured as the standard deviation of yearly income across five years. The economics literature on income has used this measure extensively, often after adjusting for life-course effects by controlling for age (Dynan et al. 2012; Hardy 2014).

To examine potentially nonlinear effects of variability, I also used a “high variability” measure, defined as greater than the median standard deviation APC. There is also reason to expect that the magnitude of a given change in income would relate to its effects on children. In keeping with prior studies, I created count variables of the number of year-to-year income changes of 25% or more (Hardy 2014) and tested the results using an alternative threshold of 50% (Dahl et al. 2011). I included separate counts of large gains and large losses to examine the effects of income insecurity (Western et al. 2016).

Income Trend

The direction and size of income growth over time were measured as the annual growth rate in the yearly family income-to-needs ratio. Income-to-needs ratios—that is, family income relative to the poverty line for family size and structure—help to account for the fact that families grow over time and require more income. The growth rate was calculated as the exponentiated coefficient minus 1 on a continuous variable capturing year in an OLS regression predicting the natural logarithm of family income-to-needs. I also used a dichotomous measure of negative or zero growth.

It is important for the conceptualization and measurement of income dynamics to note that these measures of level, variability, and growth are not highly correlated. As shown in Table 3, the correlation between level and variability is −0.22 (p < .001); the correlation between level and growth is −0.05 (p < .01); and that between variability and growth is 0.13 (p < .001). That the three dimensions are empirically distinguishable lends support to this study's attempt to distinguish them conceptually and to understand their unique influences on child development.

Child Achievement and Behavior

The dependent variables include two measures of child cognition and two measures of child behavior. The achievement measures were taken from the Woodcock-Johnson Psycho-Educational Battery-Revised (WJ-R), normed tests for measuring academic achievement. In the CDS, reading and math subtests of the WJ-R were administered to children ages 6 years or older in each wave. Children 3 to 5 years old were administered only one of four reading subtests: the letter-word identification (relevant only in Wave 1, when some sample children were under 6 years old; Duffy and Sastry 2014). The Spanish version of the WJ-R (Batería-R, Form A) was used for children whose primary language was Spanish. The normed scores were constructed based on the child's raw score on the test and the child's age to the nearest month. In this sample, the range for math (applied problems and calculation subtests in the WJ-R) was 19 to 184 points, and the mean score was 105.99 points (SD = 17.32). The range for the broad reading scores (letter-word and passage comprehension subtests) was 9 to 194 points, and the mean was 105.18 (SD = 17.29).

The behavior measures were obtained from the Behavior Problem Index (BPI) and the Positive Behavior Scale (PBS), commonly used and well-validated measures of child behavior. The BPI, which is based on Achenbach's behavior problem checklist, measures the incidence and severity of child behavior problems, such as acting out, being anxious or withdrawn, and acting hyperactive (Achenbach 1994). The BPI scale was based on responses by the primary caregiver as to whether 32 problem behaviors are often, sometimes, or never true of the targeted child. The BPI scores ranged from 0 to 30 points, with a mean of 8.28 (SD = 6.23). I also examined the results for two important subscales of the BPI: externalizing and internalizing problem behavior. The former captures disruptive behaviors, and the latter captures anxious or depressed behaviors.

The PBS measures childhood emotional and social competence (Quint et al. 1997). The original scale consisted of 25 items for children 3 years or older evaluated on a 10-point scale, ranging from “not at all like my child” to “very much like my child.” Questions ask the parent about the child being “cheerful,” “getting along with others,” and getting “over being upset quickly.” The CDS used 10 of these items, each scored from 1 to 5. In this sample, the PBS ranged from 0 to 5, with a mean of 4.17 (SD = 0.58).

Covariates and Moderators

The following time-invariant child characteristics were used from the first wave of the CDS: whether the child had low birth weight, whether the child was breastfed, the child's gender, and the child's race. I also controlled for two time-variant child characteristics, measured at each CDS wave: the child's living circumstances and the child's age in years. The child's living circumstances were measured in three mutually exclusive dichotomous variables: living with both parents, living with the mother only, and other (includes living with father only). In addition, in supplementary analyses, I controlled explicitly for family (in)stability using a binary indicator for whether the child's living circumstances were stable from one wave to the next (starting with Wave 2). More than 80% of the sample had no change in living circumstances across waves.

From the core PSID files, I controlled for characteristics of the household head, including age (continuous), marital status (married, never married, or divorced, widowed, or separated), and educational attainment (less than 12 years, 12 years, 13–15 years, or 16 or more years). Detailed wealth data were collected in certain years of the PSID: 1994, 1999, 2001, 2003, 2005, and 2007. I used the PSID constructed variables for total wealth with home equity, which can take values below 0 (i.e., debt). I dichotomized that variable into below (1) or at or above (0) median wealth ($22,720). To reduce the endogeneity between the wealth and income measures included in our models, I used measures of wealth taken in 1994 and 2001 before the income measures for Waves 2 and 3 (see Table 1). This was not possible for the first wave, so this wave was excluded from these analyses.

Results

Associations Between Income Dynamics and Child Achievement and Behavior

The bivariate associations between the measures of income dynamics and child outcome measures are shown in Table 3. As expected, income level is positively associated with test scores and negatively associated with problem behavior. Income variability is also associated with each of those outcomes in the opposite direction and at half the magnitude. The five-year trend in income-to-needs has a small bivariate association with lower test scores but no association with behavior. For the most part, the income measures are not associated with positive social behavior, with the exception of a small marginally significant and positive correlation between income trend and positive social behavior.

Table 4 presents coefficient estimates and standard errors from regression models predicting the child outcome measures as a function of continuous five-year income dynamics and a set of child- and family-level control variables. Because the income measures were all continuous in this set of models, the coefficients estimate the linear association of a one-unit increase in income level, variability, or trend on child achievement and behavior. In brackets, the beta coefficients show the associations between the income measures and the child outcome measures in standard deviation units (i.e., effect size).

Consistent with prior studies, higher average income is associated with higher math and reading test scores and with fewer problem behaviors. A $1,000 increase in income is associated with a 0.05 point (p < .001) increase in math scores, a 0.03 point (p < .001) increase in reading scores, and a 0.01 point (p < .01) decrease in problem behaviors. The coefficients on income variability are in the opposite direction of income level, but none are statistically significant. The average yearly growth rate of income is not statistically related to achievement but does predict lower average problem behavior and higher positive social behavior. These associations are small: a 1 percentage point increase in the income growth rate is associated with a 0.008 point (p < .10) reduction in problem behavior score and a 0.001 point (p < .01) increase in positive social behavior.

The direction of associations between the control variables and child development was generally expected and consistent with prior research. Average achievement is lower for non-White children, those not breastfed, and children with lower-educated parents. Some of the same characteristics are associated with adverse behavior but much less consistently. Notably, the statistically significant associations between five-year income dynamics and child achievement and behavior are much smaller in size than the associations between other child and family characteristics and those outcomes. For example, a $1,000 increase in income is associated with a 0.05 point increase on a standardized math test, but the child living in a household with a less-educated head is associated with a 7.8 point decrease in reading scores relative to higher-educated parents.

The results in Table 4 were robust to a variety of alternative specifications (results not shown), including using the standard deviation of income (transitory variance) to measure income variability, using Poisson regression for the highly skewed outcome of positive social behavior, and using total wealth as a control variable instead of homeownership. Also, in the structure of the data, there was temporal overlap between the Wave 2 income measures (1998, 2000, and 2002) and the child outcome measures (2002), raising concern about bias from simultaneity. For example, a child's behavior problems might lead to parental job loss and greater income variability. In models dropping Wave 2, the direction and statistical significance of the effects of income level remain the same, but higher standard deviation APC is related to a small marginally significant increase in problem behavior scores (0.005 points; p < .10). I also estimated the main models predicting the problem behavior externalizing and internalizing subscales separately and found that income level and trend are associated with a reduction in both types of behavior problems but that income variability does not have statistically significant associations with either.

Finally, a likely cause of income instability is a change in family composition or structure. Although this study was not focused on examining the causes of income variability, I did examine whether the relationships between income dynamics and child achievement and behavior might vary for children in (un)stable families (results not shown). I first controlled for a binary indicator of whether the child's living arrangements (i.e., which biological parents they were living with) changed from the prior wave, which dropped the first wave of data. With the inclusion of this control, the size of the coefficients on income level, variability, and trend all decreased slightly but with few differences in the statistical significance of those coefficients. The one exception was that the associations between income trend and behavior were no longer statistically significant. Limiting the sample to only children with unstable family structures also did not alter the pattern of mostly null effects.

Threshold Effects of High Variability or Low Trend

Table 5 shows the results of models using three types of categorical indicators for high thresholds of variability that would arguably be most disruptive. All these models controlled for continuous measures of income level and the standard set of control variables shown in Table 4. The results in panel A used high variability (above-median standard deviation of APC) and stagnant or downward trend (a growth rate less than or equal to 0). Panels B and C show results from models testing the hypothesis that large income gains or losses between years might be particularly disruptive to families.

The results show minimal evidence that high variability or stagnant/negative growth is associated with child development (Table 5, panel A). High income variability has a negative association with math and reading scores, but the point estimates are not statistically significant. In Panel B, the coefficients become larger (more adverse) with an increasing number of large income changes that a household experienced, but none of the coefficients are statistically different from 0, and few are statistically different from each other. The evidence on the direction of large changes is comparable (panel C), with no statistically significant associations with any outcomes. When I changed the definition of “large” income change to 50% rather than 25% (results not shown), the adverse associations between large changes and both achievement scores grew in magnitude, and two or more large income changes were negatively and significantly (p < .05) associated with reading scores. In addition, negative associations between gains only and losses and gains with reading scores were both negative and statistically significant (p < .10 and p < .05, respectively). Compared with the linear results in Table 4, these results provide little evidence that income changes might be most detrimental when larger and more frequent.

Moderating Effects of Child's Age and Race

Table 6 displays the results of the main models using continuous income dynamic measures and the full set of controls to predict child achievement and behavior by subgroups based on child's race (White versus non-White) and age (adolescents vs. nonadolescents). As noted in the Methods section, these crude groupings are necessitated by the relatively small sample of children in the PSID-CDS.

In general, the effect of income dynamics is more pronounced for non-White than White children. The associations between income level and achievement test scores are statistically significant in both groups but were larger for non-White children (p < .10). Also, the following associations are significant only for the non-White subgroup: between income level and problem behavior (−0.02; p < .001); between income variability and reading (−0.02; p < .10), problem behavior scores (0.01; p < .10), and positive social behavior (−0.00; p < .10); and between yearly income growth rate and positive social behavior (0.002; p < .01). For problem behavior and positive social behavior, the differences in the magnitudes of the coefficients on variability for Whites and non-Whites are statistically significant (shown with an “a” superscript; p < .10). Many of the associations between income dynamics and achievement are larger for younger children than for adolescents, but the coefficients are not statistically different from 0 or one another (except the association between income level and math scores). For behavior, the magnitude of the associations with income variability and trends appear identical between adolescents and nonadolescents.

Moderating Effects of Income and Wealth Levels

To test the moderating role of income and wealth levels, I predicted the child outcomes with the measures for variability and trend interacted with an indicator equal to 1 if the family's income was below the median ($54,580) or if the family's wealth (assets-debt) was below the median ($22,270). The predicted margins from these partially interacted models are shown graphically in panels a–h of both Figure 2 (income) and Figure 3 (wealth). The coefficients are shown in Table A1 of the online appendix.

The associations between income variability or trend and child achievement and behavior do not vary by median income. Notably, the association between income variability and achievement scores is negative but not statistically significant for the lower-income families. Even in cases in which the relationship looks different in Figure 2 for those with income below the median and at or above the median (e.g., the association between growth rate and reading scores), the coefficient on the interaction term is not statistically significant. The story is similar for wealth, with one exception: income growth is promotive of higher test scores only for families at or above median wealth. The relationship between income growth rate and test scores is negative when families have below-median wealth. The interaction between below-median wealth and growth rate is statistically significant for math but not for reading (Table A1).

Discussion

This study examined three dimensions of childhood income dynamics—level, trend, and variability—as predictors of child achievement and behavior. After income level was controlled for, income variability had null associations with child achievement and behavior. In contrast, income trend had modest beneficial associations with both measures of behavior but neither measure of achievement. I found modest and suggestive evidence of threshold effects, such that multiple large changes and having both gains and losses were predictive of lower test scores, on average; again, though, the point estimates were not statistically significant. The subgroup analyses found larger adverse effects of income variability for non-White than White children. The results did not differ statistically for adolescents and nonadolescents in the sample, but the coefficients were larger for the children under 13 years. Finally, contrary to expectations, there was no clear pattern of differences in the associations between income variability and trend by a family's income or wealth level.

The results of this study have three key implications. First, these findings reassert the primacy of income level in capturing the salient economic circumstances for child development. I did not find many associations between income variability or trend and child development, and the scattered associations were many times smaller than those for income level. Importantly, this result was not because income level and variability or trend could not be distinguished conceptually or empirically (i.e., because of high collinearity), a fact that adds support to the idea that income level is more relevant to parental investment and family processes than changes in income.

Why might income level be more salient? One possibility is that even though income variability is high for some families, it is not disruptive because parents are able to predict when it will occur and respond in ways that protect the developmental contexts of children (e.g., consumption of basic needs and the quality of schools). This hypothesis would suggest that income level measured as an average over multiple years captures not only a family's overall economic circumstances but also the ways in which income shapes developmental contexts—for instance, through spending and stress. Also, the mostly null or small associations between income variability and trend and child achievement and behavior might be the result of observing income only yearly (or biyearly in some waves). Some prior studies indicated that intrayear income variability is more extreme than interyear variability and that changes in weekly or monthly income may be more stressful and disruptive to family routines (Gennetian et al. 2015). Further, attrition from the longitudinal PSID is known to be higher among those with unstable earnings (Fitzgerald et al. 1998), which may mean that this analysis does not include those with the most severe income instability. In part because of these data limitations, the results of this study should not dissuade ongoing scholarship on income dynamics, family processes, and children. There is good theoretical support for the idea that income variability and trend may matter to parental stress, spending on children, and child development (Hill et al. 2013; Sandstrom and Huerta 2013). In addition, prior studies found adverse effects of income variability on adolescent and adult education outcomes (Gennetian et al. 2018b; Gennetian et al. 2015; Hardy 2014). Finally, many low-income families face multiple types of instability simultaneously or regularly, including family instability, residential instability, and income instability. Perhaps the combined effects of those disruptions are more relevant to child development.

Second, the significant and larger adverse consequences of income variability for non-White children are particularly noteworthy. It is particularly important that race—not SES—moderated the effects of income variability in this study. Black and Hispanic children are more likely than White children to experience high income variability and low or no income growth (Hardy 2014; Hill 2018) and to experience greater adverse effects of income variability on school outcomes in adolescence (Gennetian et al. 2018b; Gennetian et al. 2015; Hardy 2014). Although developmental theory has long considered the intersecting contexts of race and class (García Coll et al. 1996), studies of the effects of race and class on child development are still likely to treat them as separate individual characteristics rather than as indicators of a child's position in systems of (dis)advantage. These intersecting disadvantages can alter the benefits that income confers, how income shapes the family and other developmental contexts, and the culturally adaptive practices used to support child development (García Coll et al. 1996; Henry et al. 2019). For example, the stress associated with racial discrimination (e.g., Pascoe and Richman 2009) could leave Black and Hispanic parents and children less able to weather the disruptions associated with income variability. Knowing that income dynamics matter differently for children of different racial/ethnic identities calls for more studies that capture the variation in context by race for same-SES families (e.g., Lindsay 2011), interactions between racial and class identities (e.g., Destin et al. 2019), and the differential value of parental investments in neighborhoods and schools (Reardon et al. 2015). A key question for future researchers is whether processes related to racial identity and racial discrimination make African American and Hispanic children more vulnerable or resilient to different types of instability.

Third, two suggestive and counterintuitive findings regarding the interaction between SES and income dynamics are worth noting. First, the association between income variability and achievement scores was negative for the higher-wealth families, although the interaction was not statistically significant. Also, the association of income growth with achievement tests was positive for those with above-median income or wealth but negative for families with below-median income or wealth (only the wealth interaction on math was statistically significant). These results may relate to how low- and high-SES families spend money on children. Prior studies found that income is related to child achievement development primarily through parental investments in care, education, and basic needs (Guo and Harris 2000; Yeung et al. 2002). Also, socioeconomic inequality in parental investments have been growing over time (e.g., Kornrich and Furstenberg 2013). Perhaps higher-SES families investing disproportionally more in children are less likely to experience income variability, but it is more likely for variability to affect their children. Also plausible is that year-to-year income changes are most disruptive for more financially stable families in the same way that recent studies have found that paternal incarcerations and maternal job loss are more disruptive for the families least likely to experience them (Brand and Thomas 2014; Turney 2015).

This study has several limitations. First, the small sample of the CDS constrained the analysis of subgroups or moderated effects. For example, most evidence on age as a moderator of the effects of income on child outcomes suggests that children under age 6 are most vulnerable, but this study's sample was not large enough to examine that group specifically. Also, as an observational study, the analysis presented here assumed that all differences between children in families with different income dynamics have been observed and included in the model and that there is no reverse causal relationship by which child achievement and behavior are influencing family income dynamics. Given these strong assumptions, the results should be viewed as associational, not causal. Finally, the measurements of child achievement and behavior are not without flaws. The test scores capture achievement, which is considered a proxy for cognitive development but one that is highly affected by contextual factors (e.g., the quality of schools), and parent-reported behavior measures are known to be biased negatively if the reporter is depressed (Chi and Hinshaw 2002; Gartstein et al. 2009). Importantly, it was beyond the scope of this study to identify the causes of income variability or to examine how the effects of income dynamics differ by different family structures, employment patterns, or public assistance use. These are topics ripe for future research.

Despite these limitations, the study adds to our growing understanding of the dynamic nature of family life and particularly the salience of instability in multiple domains for lower-SES families and children. It suggests that income variability and trend are distinct features of economic circumstances during the life course, although less consequential to this set of outcomes than income level. The larger associations between income level and child achievement (compared with those between income variability and child achievement) support the focus of income support programs (e.g., Supplemental Nutrition Assistance Program and Earned Income Tax Credit) on income sufficiency and basic needs. Nonetheless, these results suggest that interventions designed to support family income or wealth could benefit children by experimenting with delivering benefits in ways that promote both sufficiency and stability. Future research should focus on the mechanisms of the effects of income variability and trend on child development, particularly whether they differ from the established mechanisms of intergenerational class transmission. Both large surveys and community studies of child development could support future research by adding higher frequency measures of family economic circumstances, parenting stress and practices, and child well-being. Finally, future studies could help to disentangle the differing effects of income variability caused by different events in family life, including parental employment instability, family structure instability, and health events.

Acknowledgments

The author thanks the Editor, Dr. Mark Hayward, and several anonymous reviewers for excellent feedback that improved this paper. In addition, the author thanks Chieko Maene at the University of Chicago, who provided research assistance during the early stages of this project, and the investigators and staff of the Panel Study of Income Dynamics at the Institute for Social Research at the University of Michigan. This research was partially supported by the Family Self-Sufficiency Research Consortium, Grant Number 90PD0290-03-00, funded by the Office of Planning, Research, and Evaluation in the Administration for Children and Families, U.S. Department of Health and Human Services; and by a Eunice Kennedy Shriver National Institute of Child Health and Human Development research infrastructure grant, P2C HD042828, to the Center for Studies in Demography & Ecology at the University of Washington. The contents of this publication are solely the responsibility of the author.

Notes

1

I omit four CDS cases from this analysis because of a known error in their family’s income variables in the 1997 and 1999 PSID (Dynan et al. 2012).

2

In the PSID, the label of “head” is assigned to the adult in the family unit with financial responsibility. In families with a male spouse or partner present, the man is automatically assigned as the head unless he is incapacitated. The spouse or long-term cohabiting partner of the head is labeled the “wife.” Other adults in the household could be short-term cohabiting partners, extended family members, or unrelated individuals. Same-gender couples in the sample receive the same labels as different-gender couples, with one exception: same-gender couples are labeled “head” and “girlfriend/boyfriend” regardless of the length of the cohabitation, and different-gender couples are labeled as “head” and “wife” after they cohabit for one year.

The text of this article is only available as a PDF.

References

Achenbach, T. M. (
1994
).
Child Behavior Checklist and related instruments
. In Maruish, M. E. (Ed.),
The use of psychological testing for treatment planning and outcome assessment
(pp.
517
549
). Retrieved from http://psycnet.apa.org/record/1994-97261-020
Becker, G. S. (
1981
).
A treatise on the family
.
Cambridge, MA
:
Harvard University Press
.
Ben-Ishai, L. (
2015
).
Volatile job schedules and access to public benefits
(Report). Available from www.clasp.org
Blau, D. M. (
1999
).
The effect of income on child development
.
Review of Economics and Statistics
,
81
,
261
276
.
Bradley, R. H., & Corwyn, R. F. (
2002
).
Socioeconomic status and child development
.
Annual Review of Psychology
,
53
,
371
399
.
Brand, J. E., & Thomas, J. S. (
2014
).
Job displacement among single mothers: Effects on children's outcomes in young adulthood
.
American Journal of Sociology
,
119
,
955
1001
.
Bronfenbrenner, U. (
1995
).
Developmental ecology through space and time: A future perspective
. In Moen, P., Elder, G. H.Jr., & Lüscher, K. (Eds.),
Examining lives in context
(pp.
619
647
). https://doi.org/10.1037/10176-018
Bronfenbrenner, U., & Morris, P. A. (
2007
).
The bioecological model of human development
. In
Handbook of child psychology
(Vol.
1
, pp.
993
1028
). https://doi.org/10.1002/9780470147658.chpsy0114
Case, A., Lubotsky, D., & Paxson, C. (
2002
).
Economic status and health in childhood: The origins of the gradient
.
American Economic Review
,
95
,
1308
1334
.
Cavanagh, S. E., & Fomby, P. (
2019
).
Family instability in the lives of American children
.
Annual Review of Sociology
,
45
,
493
513
.
Chi, T. C., & Hinshaw, S. P. (
2002
).
Mother-child relationships of children with ADHD: The role of maternal depressive symptoms and depression-related distortions
.
Journal of Abnormal Child Psychology
,
30
,
387
400
.
Coldwell, J., Pike, A., & Dunn, J. (
2006
).
Household chaos—Links with parenting and child behaviour
.
Journal of Child Psychology and Psychiatry and Allied Disciplines
,
47
,
1116
1122
.
Coley, R. L., Lynch, A. D., & Kull, M. (
2015
).
Early exposure to environmental chaos and children' s physical and mental health
.
Early Childhood Research Quarterly
,
32
,
94
104
.
Conger, R. D., Conger, K. J., Elder, G. H.Jr., Lorenz, F. O., Simons, R. L., & Whitbeck, L. B. (
1992
).
A family process model of economic hardship and adjustment of early adolescent boys
.
Child Development
,
63
,
526
541
.
Dahl, G. B., & Lochner, L. (
2012
).
The impact of family income on child achievement: Evidence from the earned income tax credit
.
American Economic Review
,
102
,
1927
1956
.
Dahl, M., De Leire, T., & Schwabish, J. A. (
2011
).
Estimates of year-to-year volatility in earnings and in household incomes from administrative, survey, and matched data
.
Journal of Human Resources
,
46
,
750
774
.
Danese, A., & McEwen, B. S. (
2012
).
Adverse childhood experiences, allostasis, allostatic load, and age-related disease
.
Physiology & Behavior
,
106
,
29
39
.
Dearing, E., McCartney, K., & Taylor, B. A. (
2001
).
Change in family income-to-needs matters more for children with less
.
Child Development
,
72
,
1779
1793
.
Dearing, E., McCartney, K., & Taylor, B. A. (
2006
).
Within-child associations between family income and externalizing and internalizing problems
.
Developmental Psychology
,
42
,
237
252
.
Destin, M., Rheinschmidt-Same, M., & Richeson, J. A. (
2019
).
Implications of intersecting socioeconomic and racial-ethnic identities for academic achievement and well-being
. In Henry, D. A., Voltruba-Drzal, E., & Miller, P. (Eds.),
Advances in child development and behavior: Child development at the instersection of race and SES
(Vol.
57
, pp.
149
164
).
Cambridge, MA
:
Elsevier Inc
.
Duffy, D., & Sastry, N. (
2014
).
Achievement tests in the Panel Study of Income Dynamics Child Development Supplement
. Retrieved from http://psidonline.isr.umich.edu/Guide/documents.aspx
Dumas, J. E., Nissley, J., Nordstrom, A., Smith, E. P., Prinz, R. J, & Levine, D. W. (
2005
).
Home chaos: Sociodemographic, parenting, interactional, and child correlates
.
Journal of Clinical Child and Adolescent Psychology
,
34
,
93
104
.
Duncan, G. J. (
1988
).
The volatility of family income over the life course
. In Baltes, P. B., Featherman, D. L., & Lerner, R. M. (Eds.),
Life-span development and behavior
(Vol.
9
, pp.
317
358
).
Hillsdale, NJ
:
Lawrence Erlbaum Associates, Publishers
.
Duncan, G. J., Kalil, A., & Ziol-Guest, K. M. (
2018
).
Parental income and children's life course: Lessons from the Panel Study of Income Dynamics
.
Annals of the American Academy of Political and Social Science
,
680
,
82
96
.
Duncan, G. J., Magnuson, K. A., Kalil, A., & Ziol-Guest, K. M. (
2012
).
The importance of early childhood poverty
.
Social Indicators Research
,
108
,
87
98
.
Dynan, K., Elmendorf, D., & Sichel, D. (
2012
).
The evolution of household income volatility
.
B.E. Journal of Economic Analysis & Policy
,
12
(
2
),
3
. https://doi.org/10.1515/1935-1682.3347
Dynarski, S., & Gruber, J. (
1997
).
Can families smooth variable earnings?
Brookings Papers on Economic Activity
,
1997
(
1
),
229
303
.
Evans, G. W., Schamberg, M. A., & McEwen, B. S. (
2009
).
Childhood poverty, chronic stress, and adult working memory
.
Proceedings of the National Academy of Sciences
,
106
,
6545
6549
.
Evans, G. W., & Wachs, T. D. (
2010
).
Chaos and its influence on children's development: An ecological perspective
.
Washington, DC
:
American Psychological Association
.
Fitzgerald, J., Gottschalk, P., & Moffitt, R. (
1998
).
An analysis of sample attrition in panel data: The Michigan Panel Study of Income Dynamics
.
Journal of Human Resources
,
33
,
251
299
.
Friedman, M. (
1957
).
The permanent income hypothesis
. In
A theory of the consumption function
(pp.
20
37
).
Princeton, NJ
:
Princeton University Press
. Retrieved from http://www.nber.org/chapters/c4405.pdf
Ganzel, B. L., & Morris, P. A. (
2011
).
Allostasis and the developing human brain: Explicit consideration of implicit models
.
Development and Psychopathology
,
23
,
955
974
.
García Coll, C., Crnic, K., Lamberty, G., Wasik, B. H., Jenkins, R., Vázquez García, H., & McAdoo, H. P. (
1996
).
An integrative model for the study of developmental competencies in minority children
.
Child Development
,
67
,
1891
1914
.
Garrett-Peters, P. T., Mokrova, I., Vernon-Feagans, L., Willoughby, M., & Pan, Y. (
2016
).
The role of household chaos in understanding relations between early poverty and children's academic achievement
.
Early Childhood Research Quarterly
,
37
,
16
25
.
Gartstein, M. A., Bridgett, D. J., Dishion, T. J., & Kaufman, N. K. (
2009
).
Depressed mood and maternal report of child behavior problems: Another look at the depression-distortion hypothesis
.
Journal of Applied Developmental Psychology
,
30
,
149
160
.
Gennetian, L. A., Castells, N., & Morris, P. A. (
2010
).
Meeting the basic needs of children: Does income matter?
Children and Youth Services Review
,
32
,
1138
1148
.
Gennetian, L. A., Rodrigues, C., Hill, H. D., & Morris, P. A. (
2018a
).
Income level and volatility by children's race and Hispanic ethnicity
.
Journal of Marriage and Family
,
81
,
204
229
.
Gennetian, L. A., Rodrigues, C., Hill, H. D., & Morris, P. A. (
2018b
).
Stability of income and school attendance among NYC students of low-income families
.
Economics of Education Review
,
63
,
20
30
.
Gennetian, L. A., Wolf, S., Hill, H. D., & Morris, P. A. (
2015
).
Intrayear household income dynamics and adolescent school behavior
.
Demography
,
52
,
455
483
.
Gottschalk, P., & Moffitt, R. A. (
2009
).
The rising instability of U.S. earnings
.
Journal of Economic Perspectives
,
23
(
4
),
3
24
.
Guo, G., & Harris, K. M. (
2000
).
The mechanisms mediating the effects of poverty on children's intellectual development
.
Demography
,
37
,
431
447
.
Ha, Y., Thomas, M. M. C., Byrne, T., & Miller, D. P. (
2020
).
Patterns of multiple instability among low-income families with children
.
Social Service Review
,
94
,
129
168
.
Hacker, J. S. (
2008
).
The great risk shift: The new economic insecurity and the decline of the American Dream
.
Oxford, UK
:
Oxford University Press
.
Hamilton, D., & Darity, W. A. (
2017
).
The political economy of education, financial literacy, and the racial wealth gap
.
Federal Reserve Bank of St. Louis Review
,
99
,
59
76
.
Hardy, B. L. (
2012
).
Black female earnings and income volatility
.
Review of Black Political Economy
,
39
,
465
475
.
Hardy, B. L. (
2014
).
Childhood income volatility and adult outcomes
.
Demography
,
51
,
1641
1665
.
Hardy, B. L. (
2017
).
Income instability and the response of the safety net
.
Contemporary Economic Policy
,
35
,
312
330
.
Hardy, B. L., & Ziliak, J. P. (
2014
).
Decomposing trends in income volatility: The “wild ride” at the top and bottom
.
Economic Inquiry
,
52
,
459
476
.
Heckman, J. J. (
2006
).
Skill formation and the economics of investing in disadvantaged children
.
Science
,
312
,
1900
1902
.
Henry, D. A., Votruba-Drzal, E., & Miller, P. (
2019
).
Child development at the intersection of race and SES: An overview
. In Henry, D. A., Votruba-Drzal, E., & Miller, P. (Eds.),
Advances in child development and behavior: Child development at the intersection of race and SES
(Vol.
57
, pp.
1
25
).
Cambridge, MA
:
Elsevier
.
Hill, H. D. (
2018
).
Trends and divergences in childhood income dynamics: 1970s–2000s
. In Benson, J. B. (Ed.),
Advances in child development and behavior
(Vol.
54
, pp.
179
213
).
Cambridge, MA
:
Elsevier
.
Hill, H. D., Morris, P. A., Gennetian, L. A., Wolf, S., & Tubbs, C. Y. (
2013
).
The consequences of income instability for children's well-being
.
Child Development Perspectives
,
7
,
85
90
.
Hill, H. D., Romich, J., Mattingly, M. J., Shamsuddin, S., & Wething, H. (
2017
).
An introduction to household economic instability and social policy
.
Social Service Review
,
91
,
371
389
.
Institute for Social Research
. (n.d.).
Panel Study of Income Dynamics
. Retrieved from https://psidonline.isr.umich.edu/studies.aspx
Kalleberg, A. L. (
2010
).
Good jobs, bad jobs: The rise of polarized and precarious employment systems in the United States, 1970s to 2000s
.
New York, NY
:
Russell Sage Foundation
.
Kornrich, S., & Furstenberg, F. (
2013
).
Investing in children: Changes in parental spending on children, 1972–2007
.
Demography
,
50
,
1
23
.
Lichter, D. T., & Eggebeen, D. J. (
1993
).
Rich kids, poor kids: Changing income inequality among American children
.
Social Forces
,
71
,
761
780
.
Lindsay, C. A. (
2011
).
All middle-class families are not created equal: Explaining the contexts that Black and White families face and the implications for adolescent achievement
.
Social Science Quarterly
,
92
,
761
781
.
McKernan, S.-M., Ratcliffe, C., & Zhang, S. (
2013
).
How have wealth inequality and income inequality changed over time?
(Urban Institute report). Retrieved from http://www.urban.org/sites/default/files/alfresco/publication-pdfs/412802-less-than-equal-racial-disparities-in-wealth-accumulation.pdf
McLanahan, S. (
2004
).
Diverging destinies: How children are faring under the Second Demographic Transition
.
Demography
,
41
,
607
627
.
McLoyd, V. C. (
1990
).
The impact of economic hardship on Black families and children: Psychological distress, parenting, and socioemotional development
.
Child Development
,
61
,
311
346
.
Miller, G. E., Chen, E., & Parker, K. J. (
2011
).
Psychological stress in childhood and susceptibility to the chronic diseases of aging: Moving toward a model of behavioral and biological mechanisms
.
Psychological Bulletin
,
137
,
959
997
.
Mistry, R. S., Benner, A. D., Tan, C. S., & Kim, S. Y. (
2009
).
Family economic stress and academic well-being among Chinese-American youth: The influence of adolescents' perceptions of economic strain
.
Journal of Family Psychology
,
23
,
279
290
.
Pascoe, E. A., & Richman, L. S. (
2009
).
Perceived discrimination and health: A meta-analytic review
.
Psychological Bulletin
,
135
,
531
554
.
Perkins, K. L. (
2017
).
Household complexity and change among children in the United States, 1984 to 2010
.
Sociological Science
,
4
,
701
724
.
Quint, J. C., Bos, J. M., & Polit, D. F. (
1997
).
New chance: Final report on a comprehensive program for young mothers in poverty and their children
. Retrieved from https://files.eric.ed.gov/fulltext/ED419864.pdf
Ratcliffe, C. (
2015
).
Child poverty and adult success
(Urban Institute Low-Income Working Families Initiative report). Retrieved from https://www.urban.org/sites/default/files/publication/65766/2000369-Child-Poverty-and-Adult-Success.pdf
Reardon, S. F., Fox, L., & Townsend, J. (
2015
).
Neighborhood income composition by household race and income, 1990–2009
.
Annals of the American Academy of Political and Social Science
,
660
,
78
97
.
Sandoval, D. A. R. (
2009
).
The increasing risk of poverty across the American life course
.
Demography
,
4
,
717
738
.
Sandstrom, H., & Huerta, S. (
2013
).
The negative effects of instability on child development: A research synthesis
(Urban Institute Low-Income Working Families Fact Sheet). Retrieved from https://www.urban.org/sites/default/files/publication/32721/412908-the-negative-effects-of-instability-on-child-development-fact-sheet.pdf
Shonkoff, J. P., Garner, A. S., &
Committee on Psychosocial Aspects of Child and Family Health
. (
2012
).
The lifelong effects of early childhood adversity and toxic stress
.
Pediatrics
,
129
,
e232
e246
. https://doi.org/10.1542/peds.2011-2663
Turney, K. (
2015
).
Beyond average effects: Incorporating heterogeneous treatment effects into family research
.
Journal of Family Theory & Review
,
7
,
468
481
.
Turney, K., & Halpern-Meekin, S. (
2017
).
Parenting in on/off relationships: The link between relationship churning and father involvement
.
Demography
,
54
,
861
886
.
Vernon-Feagans, L., Willoughby, M., & Garrett-Peters, P. (
2016
).
Predictors of behavioral regulation in kindergarten: Household chaos, parenting, and early executive functions
.
Developmental Psychology
,
52
,
430
441
.
Western, B., Bloome, D., Sosnaud, B., & Tach, L. (
2012
).
Economic insecurity and social stratification
.
Annual Review of Sociology
,
38
,
341
359
.
Western, B., Bloome, D., Sosnaud, B., & Tach, L. M. (
2016
).
Trends in income insecurity among U.S. children, 1984–2010
.
Demography
,
53
,
419
447
.
Yeung, W. J. J., Linver, M. R., & Brooks-Gunn, J. (
2002
).
How money matters for young children's development: Parental investment and family processes
.
Child Development
,
73
,
1861
1879
.
This is an open access article distributed under the terms of a Creative Commons license (CC BY-NC-ND 4.0).