In light of the committee’s charge to identify programs that would reduce child poverty in the United States by half within a decade, and to set the stage for the program and policy proposals we make later in this report, in this chapter we provide an overview of child poverty in the United States. We begin with a brief explanation of how poverty is defined. Next we offer an overview showing which demographic subgroups of U.S. children suffer the highest poverty rates today and how child poverty rates have changed over time. The chapter’s final section compares the extent of child poverty in the United States and in peer English-speaking countries. The impacts of poverty on child development are discussed in Chapter 3, while contextual factors that reinforce poverty among low-income families are discussed in Chapter 8.
“Poverty” typically refers to a lack of economic resources, but measuring it requires careful consideration of the types of economic resources to be counted as well as agreement on a minimum threshold below which a family’s economic resources may be considered insufficient. In the 1960s, the U.S. federal government developed a method for identifying a threshold amount of household cash income below which a given household, and all related individuals living in that household, would be designated as “poor.” (See Appendix D, 2-1 for a brief history of poverty measurement in the United States.) This Official Poverty Measure (OPM) of income poverty is still being used to determine social program eligibility as well as to track
long-term trends in poverty rates. Also available are poverty measures based on consumption instead of income. Nevertheless, the statement of task for our committee directed us to use the Supplemental Poverty Measure (SPM) of income poverty, which we adjusted for underreporting of some types of income in the survey data. Box 2-1 illustrates differences in estimated child poverty among these measures. For the reasons detailed in Appendix D, 2-2 (on income poverty) and Appendix D, 2-3 (on consumption poverty),
we consider the adjusted SPM to be currently the best available approach to poverty measurement.1
Measuring Poverty with the Supplemental Poverty Measure
For this report, the committee was directed to use the SPM, which bases poverty thresholds on the expenditures U.S. families must make for food, clothing, shelter, and utilities (FCSU) plus a small additional amount for other needs (such as personal care, transportation, and household supplies). Expenditures are measured using the average of 5 years of data from the Consumer Expenditure Survey, with the poverty threshold set at the level of FCSU expenditures for family units with two children, which separates the bottom one-third of such families, ranked by FCSU expenditures, from the top two-thirds. For 2016, thresholds ranged between about $22,000 and $26,000 for two-adult, two-child families, depending on whether the family owned or rented its housing (Fox, 2017). The SPM thresholds are also adjusted for family size, using an equivalence scale, and for local cost-of-living differences in housing.2
The household resources considered are the sum of money income from all sources, including earnings and government cash benefits such as Social Security and Unemployment Compensation. A key difference between the OPM and SPM is that SPM-based household resources also include “near-cash” income benefits such as the Supplemental Nutrition Assistance Program (SNAP, formerly called food stamps) and housing subsidies, as well as benefits from many smaller programs. Deducted from household resources are child care and other work expenses, child support payments made, and out-of-pocket medical expenses (including insurance premiums).
Taxes paid, most notably payroll taxes, are also deducted from household resources, while refundable tax credits from programs like the Earned Income Tax Credit, Child Tax Credit, and the Additional Child Tax Credit are added to resources. As we show below, because government spending on tax credits and programs that provide “near-cash” (as opposed to cash) benefits have grown markedly over the past 50 years, conclusions about trends in child poverty largely depend on whether poverty is measured using the OPM or the SPM. Key differences between the official measure and the SPM are summarized in Table 2-1 and in Appendix D, 2-6.
1 The large literature of poverty measurement, in the United States and abroad, addresses types of poverty measures and measurement issues that are not central to our charge—for example, the merits of deprivation indexes compared with income- or consumption-based indexes. We briefly note these other measures and measurement issues in Appendix D, 2-2.
2Appendix D, 2-4 provides a detailed explanation of how equivalence scales are used to adjust threshold levels. Appendix D, 2-5 discusses how cost-of-living adjustments (COLAs) are used in the SPM, including how geographic COLAs compensate for differences in the price of rental housing.
TABLE 2-1 Key Differences in Poverty Measure Concepts Between the Official Poverty Measure (OPM) and the Supplemental Poverty Measure (SPM)
|Concept||Official Poverty Measure||Supplemental Poverty Measure|
|Measurement Units||Families (individuals related by birth, marriage, or adoption) or unrelated individuals||Resource units (official family definition plus any co-resident unrelated children, foster children, or unmarried partners and their relatives) or unrelated individuals (who are not otherwise included in the family definition)|
|Poverty Threshold||Three times the cost of a minimum food diet in 1963||Based on expenditures for food, clothing, shelter, and utilities (FCSU), and a little more|
|Threshold Adjustments||Vary by family size, composition, and age of householder||Vary by family size and composition, as well as geographic adjustments for differences in housing costs by tenure|
|Updating Thresholds||Consumer Price Index: all items||5-year moving average of expenditures on FCSU|
|Resource Measure||Gross before-tax cash income||Sum of cash income, plus noncash benefits that resource units can use to meet their FCSU needs, minus taxes (or plus tax credits), minus work expenses, out-of-pocket medical expenses, and child support paid to another household|
SOURCE: Fox (2017).
The Census Bureau has published SPM-based poverty statistics every fall since 2011. Its most recent report (Fox, 2018) indicates that, in 2017, 15.6 percent of children lived in families with incomes below the SPM-based poverty line. That rate is lower than the 18.0 percent rate based on the OPM (Semega, Fontenat, and Kollar, 2017), owing primarily to the SPM’s more comprehensive measure of household resources. For certain demographic groups other than children, poverty rates are higher when measured by the SPM as compared with the OPM. An example is the elderly, whose higher out-of-pocket medical payments are deducted from household resources in the SPM but not in the OPM. In addition, the 15.6 percent overall child poverty rate conceals considerable demographic and geographic variation, which we explore in subsequent sections of this chapter and Appendix D, 2-8 and 2-9.
The committee’s statement of task directs it to identify programs and policies that reduce both SPM-based poverty and deep poverty by half in
10 years. To address deep poverty, the committee adopted a common definition, namely having resources below 50 percent of those used to define poverty based on the SPM. We also provide data on “near poor” children by including those with household resources below 150 percent of poverty. These three sets of thresholds are used consistently throughout this report.
Indirect Treatment of Health Care Needs and Benefits in the SPM
One important family need that is difficult to incorporate into poverty measurement is health care—both households’ medical costs and the extent to which health insurance programs for low-income families help households afford them and shield families from falling into poverty as a result of health shocks. The importance of health care and health insurance has historically been recognized by making health insurance through Medicaid part of the package of benefits offered to low-income families such as those who qualified for Aid to Families with Dependent Children (the program that preceded Temporary Assistance for Needy Families [TANF]).
The OPM takes no account of health care needs or insurance benefits. It was developed before the life-extending advances of the past 50 years in medical treatments, such as treatment for childhood cancer, and before the expansion of health insurance to cover such treatments. However, for reasons detailed in National Research Council (1995; see also Remler, Korenman, and Hyson, 2017, and the discussion in Chapter 7), the SPM takes only a partial step forward. SPM thresholds do not include any estimated expenditure amounts for medical care, but the SPM definition of resources subtracts families’ medical out-of-pocket expenditures for any insurance premiums, copayments, deductibles, or bills for uncovered care.3 This deduction of medical out-of-pocket expenses puts some people below the SPM poverty line whom the OPM would not count as poor.4 Conversely, reductions in out-of-pocket medical care costs—through Medicaid expansion, for example—will reduce measured SPM poverty rates, all else equal (see, e.g., Summers and Oellerich, 2013).
These adjustments in the SPM, despite being a step forward, still do not account for the full contribution of government health insurance programs to reducing poverty, particularly Medicaid and the Children’s Health Insurance Program (CHIP), to the well-being of low-income
3 The reason for subtracting medical out-of-pocket costs is that, unless low-income families receive free care from providers or qualify for insurance (e.g., Medicaid) that does not require the family to contribute toward their care, then obtaining health care will require out-of-pocket expenditures.
4 For example, see U.S. Census Bureau, Table A-6: Effect of Individual Elements on SPM Poverty Rates: 2016 and 2015, September 21, 2017. Available: https://www.census.gov/library/publications/2017/demo/p60-261.html.
children and their parents. As we discuss in more detail in Chapter 7, one problem with the SPM approach is that families that defer medical care because they cannot afford it will appear to be better off than they really are. On the other hand, families who are covered by Medicaid but have little or no out-of-pocket expenses in a particular year will be appear to be worse off than they really are, because having insurance in case of future illness is much better than having no insurance at all. Nevertheless, both types of families are treated the same in this instance by the SPM. As we discuss in Chapter 7, a conceptually complete approach to the problem, one suggested in a paper by Korenman, Remler, and Hyson (2017) commissioned by the committee is to include the value of a basic health insurance plan in the poverty threshold and to include in resources the amount of government subsidy received by a family for insurance coverage. Korenman, Remler, and Hyson report some new estimates of the impact of Medicaid on poverty using this approach (see Chapter 7).5
Adjusting the Supplemental Poverty Measure Using the TRIM3 Model
Both the SPM and the OPM poverty rates are based on annual data from government surveys. To obtain these data, large national samples of households are chosen at random to participate in the Annual Social and Economic Supplement (ASEC) to the Current Population Survey (CPS), which is conducted by the U.S. Bureau of Labor Statistics. Consequently, both poverty rates are subject to bias when households misreport their incomes. The total amount of income that households report receiving from social programs in a given year can be checked against estimates of the total benefits that were paid out based on government administrative data. These comparisons often reveal large discrepancies, which have grown over time (Meyer, Mok, and Sullivan, 2009; Moffitt and Scholz, 2009; Wheaton, 2008). For example, household reports of food stamp income in the 1986 CPS accounted for 71 percent of administrative benefit totals, but in the 2006 CPS they accounted for only 54 percent of administrative benefit totals (Meyer, Mok, and Sullivan, 2009).
5 We discuss the benefits of Medicaid and CHIP in improving child health in Chapter 3. An alternative approach to valuing health care for poor families is to create a medical care financial risk index; this is discussed in Institute of Medicine and National Research Council (2012). This is a useful perspective and adds extra information to how risk varies by income in the population, but it not easily incorporated into a poverty index (Korenman, Remler, and Hyson, 2018).
To address this underreporting, the committee relied on the Transfer Income Model, Version 3 (TRIM3).6 TRIM3 is a microsimulation model that adjusts benefits from tax and transfer programs across households so that aggregated benefits reported by or assigned to households match the totals shown by administrative records.7 Imputing or modeling government transfers in this manner increases the estimated incomes of many low-income households, and in some cases it raises them above a poverty threshold. As a result, the SPM-based child poverty rates presented in this chapter and used in the policy simulations in Chapters 4, 5, and 6 are almost always lower than SPM rates reported in Census Bureau publications.
The committee used the most recent version of the TRIM3 model that was available when the bulk of its simulation work was conducted. It is based on incomes in calendar year 2015 as reported in the 2016 ASEC. Importantly, that version of TRIM3 is based on program rules and federal and state tax codes that prevailed in 2015.8 Given the potential importance of changes in federal income tax rules taking effect in 2018, the committee includes some data in later chapters showing that its key conclusions regarding child poverty reductions associated with program and policy proposals were largely unaffected by the recent changes in the tax code.
Figure 2-1 compares child poverty rates using the OPM and SPM, as well as using our modification of the SPM—labeled “TRIM3-SPM” in the remainder of this report—which is adjusted for underreported income. Some of the differences are stark. Based on the conventional definition of OPM poverty (household income below 100% of the applicable poverty line, with no adjustment for underreporting of income), nearly one-fifth (19.7%) of U.S. children—14.5 million children in all—were poor in 2015.9 The addition of tax credits, in-kind income, and other adjustments in the
6 TRIM3 is developed and managed by the Urban Institute with primary funding from the U.S. Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation. See http://trim3.urban.org/T3Welcome.php for more details about the TRIM3 model.
7 TRIM3 corrects underreporting of TANF, SSI, and SNAP only. In the 2001 CPS, just 52% of self-employment income was reported, 59% of dividends, 70% of retirement and disability benefits (excluding Social Security and Workers’ Compensation), and 73% of interest income. Unemployment Compensation is also underreported and not corrected by TRIM3. Discussions of these and other estimates are provided in Winship (2016, Appendix 3). In contrast, earnings are actually overreported at the bottom of the CPS earnings distribution when compared to administrative data (Bollinger et al., 2018; Hokayem, Bollinger, and Ziliak, 2015).
8 TRIM3 baselines for a particular year always involve applying that year’s rules to that year’s data. The results are aligned and validated using the actual benefit and tax data for that year.
9 The 19.7% figure for 2015 SPM-based poverty is considerably higher than the 18.0% figure reported above in Fox (2017), because the latter is based on 2016 incomes. Economic growth between 2015 and 2016 increased family income and decreased poverty rates among low-income families.
SPM drove the poverty rate down to 16.3 percent. Census Bureau publications use the “SPM, no adjustment for underreporting” poverty measure in their reports. But adjustments for underreporting reduced the SPM child poverty rate to 13.0 percent. Such large impacts from adjusting poverty rate estimates for underreporting of income—a 3.3 percentage-point reduction in the case of child poverty in 2015—led the committee to one of its research recommendations, presented in Chapter 9.
Although it produces a poverty count that is one-third lower than the official OPM-based count reported by the Census Bureau, our adjusted SPM-based poverty rate of 13.0 percent still represents 9.6 million U.S. children living in households with economic resources judged by the SPM
to be inadequate. The congressional charge to the committee is to identify programs that—either alone or in combination—would lift nearly 5 million of these 9.6 million children out of poverty within 10 years.
A look at rates of deep poverty, defined by the percentage of children whose families’ resource levels are less than half the poverty line, shows even more measurement sensitivity to the inclusion of taxes, in-kind income, and adjustments for underreporting. According to the OPM, which makes none of those adjustments, some 8.9 percent of children lived in deep poverty in 2015. When all three adjustments are made, the deep-poverty rate drops by more than two-thirds, to 2.9 percent. This 2.9 percent rate translates into 2.1 million children living in households with grossly inadequate resources. The congressional charge to the committee regarding deep poverty is identifying programs and policies that reduce this 2.1 million figure by more than 1 million children.
By contrast, when poverty is defined to include the “near poor”—those with incomes up to 150 percent of the poverty line—the 31.4 percent rate based on the OPM actually increases: It rises to 38.1 percent with no adjustments for underreporting and to 35.6 percent with adjustments. Substantial numbers of near-poor families pay more in taxes than they receive in tax credits, and they also incur additional work-related expenses. These factors combine to reduce net incomes and push some near-poor families below 150 percent of the SPM poverty line (Short and Smeeding, 2012).
CONCLUSION 2-1: The Supplemental Poverty Measure (SPM) has advantages over the Official Poverty Measure (OPM), the most important of which is that it includes government benefits, such as near-cash benefits and tax transfers, which are not included in the OPM. Current estimates of child poverty based on the SPM are substantially lower than those based on the OPM, and lower still when the SPM poverty estimate is adjusted for the underreporting of income in Census Bureau surveys. SPM-based estimates of poverty, combined with underreporting adjustments, indicate that 13.0 percent of U.S. children—more than 9.6 million children in all—were poor in 2015. In the case of deep poverty (defined by 50% of the SPM poverty thresholds), the corresponding rate is 2.9 percent, representing 2.1 million children.
Policy makers and researchers share a broad interest in understanding the distribution of poverty as well as the impacts of poverty-reducing programs across demographic groups. In this section, we therefore discuss how child poverty varies according to six demographic factors: race and
ethnicity, maternal schooling, family structure, adult work, immigration status, and parent’s age. Throughout this section, except where defined otherwise, the poverty rates we cite are based on the TRIM3-SPM measure described in the previous sections.
Note that a complete set of poverty-rate estimates for selected demographic groups and definitions, provided in Appendix D, Table D2-5 and Appendix E, includes demographic breakdowns not discussed in this chapter. Also note that American Indian and Alaska Native status is not included because the ASEC data did not provide a sufficient sample size to support reliable estimates for this group; a discussion of American Indian and Alaska Native child poverty using other sources of data is provided in Appendix D, 2-7 and in a research recommendation in Chapter 9.
Race and Ethnicity
The U.S. population is becoming more racially and ethnically diverse, and the diversity of the child population is increasing even more rapidly than that of the population as a whole. As detailed in Appendix D, 2-8, the proportion of racial/ethnic minority children in the total U.S. child population increased from less than one-third in 1990 to nearly one-half in 2017 (U.S. Census Bureau, 2018). The Hispanic child population has shown especially dramatic growth, increasing from 9 percent in 1980 to 25 percent in 2017 (U.S. Census Bureau, 2018). According to the Census Bureau, as of 2013 racial/ethnic minority groups combined comprised more than 50 percent of the population of children under age 1 (Pew Research Center, 2016). By 2020, the entire child population is projected to include more Hispanics, Blacks, Asians, and other minorities than non-Hispanic Whites (U.S. Census Bureau, 2018).
Concerns over varying rates of child poverty across racial/ethnic groups are long-standing (Eggebeen and Lichter, 1991; Hill, 2018; Lichter, Qian, and Crowley, 2008). These differences are readily apparent in our TRIM3-SPM-based estimates, as shown in Figure 2-2. Hispanic children experience the highest rates of poverty and deep poverty. The poverty rates for Black (17.8%) and Hispanic (21.7%) children were more than double those of non-Hispanic White (7.9%) children.10 Similar relative disparities are found for rates of deep poverty. If the line is drawn at 150 percent of SPM to include near poverty, more than one-half of all Black (50.6%) and Hispanic (54.6%) children, but less than one in four (22.9%) non-Hispanic White children, are counted as poor or near poor.
10 The TRIM3-SPM poverty rate for children in the Other Races (non-Hispanic) category, which includes American Indian and Alaska Native, Asian and Pacific Islander, and multiracial children, is 11.1%.
Another way of describing poverty across racial/ethnic groups is by asking what share of a given poverty group comprises children from specific racial/ethnic categories. Such a breakdown of data is shown in Figure 2-3.11 Again using our TRIM3-SPM-based estimates, non-Hispanic White children comprise a little more than one-half of all children but only about one-third of children in poverty or in deep poverty. The largest share of poor children are Hispanic. Similar shares of children in deep poverty are Hispanic and non-Hispanic White.
CONCLUSION 2-2: Poverty rates for children vary greatly by the child’s race and ethnicity. Based on our Transfer Income Model, Version 3 Supplemental Poverty Measure poverty estimates, Black and Hispanic children have substantially higher rates of poverty and deep poverty than non-Hispanic White children. Hispanic children constitute the largest share of poor children and nearly as large a share of deeply poor children as non-Hispanic Whites.
Education of Parents
Adults’ educational attainment is a strong correlate of their poverty status (National Academies of Sciences, Engineering, and Medicine, 2017; Wood, 2003). Completing more schooling is associated with higher rates of employment, higher earnings, better health, and a greater chance of having a spouse or partner, all of which are in turn associated with higher household income (Child Trends Data Bank, 2016). Figure 2-4 shows that child poverty rates are inversely related to the education level of the parents. Based on the TRIM3 model, one-third of children whose parents dropped
out of high school are living below the 100 percent SPM poverty line and more than two-thirds (70.7%) of these children are within 150 percent of the SPM poverty line.
Family structure has grown increasingly diverse over recent decades (Furstenberg, 2014); for example, more than 40 percent of children today are born to unmarried parents (Martin et al., 2018) and more than one-half of children will spend some of their childhood not living with both of their biological parents (McLanahan and Jencks, 2015). Although most unmarried biological parents are living together when their child is born, nearly half of these couples will separate before that child’s 5th birthday (Kennedy and Bumpass, 2008). Children born to unmarried parents may experience several different family structures over the course of their childhoods, such as living with a step-parent, with a grandparent, or in single-parent households (Manning, Brown, and Stykes, 2014). The proportion of children in single female-headed households is substantially higher for Black children (57%) than for either White (18%) or Hispanic (32%) children (National Center for Education Statistics, 2018).
For children living with both biological parents, our TRIM3 estimates find that poverty rates are less than one-half those of children with other family structures (see Figure 2-5). But even given the economic advantages of having two potential earners in the household, more than one in four (27.5%) children living with their two biological parents have family incomes below the 150 percent (near-poor) poverty line. Children living with a single parent or with neither biological parent (including foster children) have the highest rates of poverty and deep poverty.
Workers in the Household
Nearly four-fifths of all children live in families with at least one full-time working adult and, as shown in Figure 2-6, the TRIM3 SPM poverty rates for these children (6.5%) are correspondingly low. The poverty rates among children living with a part-time, as opposed to full-time, worker are correspondingly higher. By far the highest child poverty rates are observed for the relatively small fraction (6.3%) of children living in households with no workers: nearly one-quarter (22.3%) of these children are in deep poverty, three-fifths (61.5%) are below the poverty line, and the vast majority (90.8%) are below the 150 percent near-poverty line.
Children in immigrant families, defined as those with at least one foreign-born parent, represent about one-quarter of all children (Woods and Hanson, 2016).12 The TRIM3 SPM poverty rate of children in immigrant families (20.9%) is twice as high as that of children in nonimmigrant families (9.9%) (Appendix D, Table D2-6). The majority of children in immigrant families are U.S. citizens: Some 88 percent of all children in all types of immigrant households are citizens, and 79 percent of children living in households with members who are unauthorized immigrants are citizens. The immigrant status of their families is associated with a higher risk of poverty (Capps, Fix, and Zong, 2016; Migration Policy Institute, 2017; Woods and Hanson, 2016).
The relationship between poverty, citizenship, and immigration status is shown in Figure 2-7 and Appendix D, Table D2-6, again based on the TRIM3-SPM model. Children living in households in which all members are citizens have a poverty rate of 10.2 percent, nearly 3 percentage points below the 13.0 percent overall child poverty rate. By contrast, living in households with noncitizens—particularly unauthorized immigrants—is associated with higher poverty rates, even for children who are themselves U.S. citizens.
When the household includes recent or unauthorized immigrants, the poverty rate among noncitizen children is even higher: 31.8 percent and 33.3 percent, respectively. Citizenship for the child appears to buy very little in the way of poverty reduction if other household members are unauthorized: 31.5 percent of citizen children whose households have at least one unauthorized resident are poor, as are 24.7 percent of citizen children whose households have at least one recent immigrant. However, child citizenship is associated with a much lower rate of deep poverty: 6.4 percent versus 15.2 percent, respectively, for citizen versus noncitizen children, in both cases living with unauthorized household members.
12 In the TRIM3 analyses, a child is considered to have an immigrant parent if he or she has at least one biological, adoptive, or stepparent that was born in another country. A recent immigrant is defined as a person entering as a legal permanent resident within the last 5 years. Children are classified by their own status. For example, in the case of an SPM unit containing unauthorized immigrant parents, an unauthorized immigrant child, and a native-born citizen child, the unauthorized immigrant child would be categorized as “Child is a noncitizen, unit contains unauthorized immigrant” and the native-born child would be classified as “Child is a citizen, unit contains unauthorized immigrant.”
Age of Parent
Our final demographic dimension is the age of the parent, defined as the age of the biological parent, adoptive parent, or stepparent if present.13 Children born to younger mothers are more likely to live in poverty (Mather, 2010). On average, maternal age at first birth has been increasing (Mathews and Hamilton, 2016), and over the last three decades births to teen
13 Age of parent is determined first by asking the mother, if present. If the mother is not present, then the biological, adoptive, or stepfather (if present) is asked.
mothers have declined very significantly—by more than 64 percent (Martin, Hamilton, and Osterman, 2017). Despite these trends, in 2015 more than one-quarter of children were born to mothers under age 25, and racial/ethnic minority children were more likely than their White counterparts to be born to young mothers (Martin, Hamilton, and Osterman, 2017).
The poverty risk for living with a younger parent (which we define here as under age 25) is readily apparent in Figure 2-8; nearly one-quarter (23.8%) of children living with a young parent fall below the 100-percent-of-SPM
poverty line.14 Nearly three-fifths of children with a young parent live in families with incomes less than 150 percent of the poverty line.
CONCLUSION 2-3: Poverty rates for children vary greatly depending on other characteristics of parents and households. Higher poverty rates are associated with low levels of parental schooling and with living with a single parent, no parent, or a young parent. Poverty is more prevalent when both children and other family members are not citizens, although these poverty rates improve only a little when children are U.S. citizens but living in households with family members who are unauthorized. Children in families with no workers have by far the highest rates of poverty and near poverty, but even full-time work is insufficient to lift one-quarter of children living with full-time workers above the 150 percent Supplemental Poverty Measure poverty line.
Geographic Distribution of Poverty
Child poverty rates also vary across communities. As documented in Chapter 8, the experience of child poverty in a community with good schools, resources for families, and pathways for economic mobility may be different than the experience in a community that has suffered from persistent poverty for decades.
To examine the geographic distribution of both point-in-time and persistent poverty, we use county data based on the OPM, because SPM county-level estimates are not available (see Appendix D, 2-9).15 For the point-in-time analyses, we classified counties as poor if 20 percent or more of children under age 18 lived in families with incomes below poverty thresholds in 2015. As shown in Figure 2-9 and Appendix D, 2-9, nearly all counties in the South and Southwest and many counties in the West and the Appalachian region had child poverty rates of 20 percent or higher in 2015. Relative to the total number of children of a given race and ethnicity, the
14 Note this is not the age at birth but the age of the parent at the time of the survey. As shown in the notes to Figure 2.8, only 4.5% of all parents of children less than 18 are to parents of age less than 25.
15 The committee assessed the lowest geographic disaggregation level that can be achieved with the SPM and found that there are no county or other substate (besides metropolitan area) SPM estimates. This is primarily because the CPS ASEC is the primary dataset used for SPM, and its sample size does not allow estimates for such small geographic areas. Because of its larger sample size, the ACS is the most likely alternate dataset, but it is missing critical variables used in calculating SPM. While there has been some work, primarily Renwick (2015), has experimented with using the CPS ASEC to inform ACS imputations of missing variables so that the ACS can hypothetically be used to estimate substate SPM; in the end those researchers created only state-level (single-year) estimates and reached no conclusions about substate level SPM estimates.
risk of residing in a point-in-time poor county is highest among Black children (70.8%), followed by American Indian and Alaskan Native (70.6%), Hispanic (65.0%), and non-Hispanic White children (46%).
We also examined the geographic distribution of persistently high child poverty. A county was classified as having persistently high child poverty if 20 percent or more of its children were classified as OPM-poor over four decades: in the 1980, 1990, and 2000 decennial censuses and in the 2007–2011 American Community Survey 5-year estimates (see Appendix D, 2-9). Some 10.2 million children (13.9% of all children) lived in persistently poor counties in 2015. The 10.2 million figure includes 3.6 million White children, 3.1 million Hispanic children and 2.7 million Black children (refer to Figure 2-3). The risk of living in a persistently poor county is highest among American Indian and Alaska Native children (36%) followed by Black (27%), Hispanic (17.1%), non-Hispanic White (9.4%), and Asian and Pacific Islander (8.2%) children (Appendix D, Figure D2-7).
Persistently high poverty is more geographically concentrated than point-in-time poverty (see Figure 2-10). The South and Northeast regions have the highest proportion of children in persistently poor counties (22.1% and 17.3%, respectively; see Appendix D, Figure D2-9) and account for the vast majority of children (81.3%) living in those counties. Although not readily apparent in Figure 2-10, due to their small land mass the persistently poor counties in the Northeast, which include the cities of New York, Philadelphia, Newark, and Boston, account for 2.1 million children.
CONCLUSION 2-4: Poverty rates for children vary considerably by geographic location. About one in seven children live in counties with persistently high child poverty (Official Poverty Measure child poverty rates always above 20% since 1980). The South and several large metropolitan areas in the Northeast regions have the highest proportions of children in counties with persistently high child poverty.
Historical trends in the OPM are published annually by the Census Bureau. As shown in Figure 2-11, they suggest that virtually no progress has been made in reducing child poverty between the late 1960s and today. If anything, child poverty rates as measured by the OPM were a little higher in 2016 (18.0%) than they had been 50 years before, in 1967 (16.6%; U.S. Census Bureau, 2018, Table 3).
Given the growth in near-cash benefits over this period, it is possible that child poverty rates based on the SPM, which counts most near-cash benefits as income, and the OPM, which does not, may show different trends. A first step in investigating whether this is the case is to construct a consistent time series of SPM-based rates, as shown in Figure 2-11 (Hardy, Smeeding, and Ziliak, 2018).
Two complications arise. First, because some TRIM3 adjustments are not available for most of the years we examined, the analyses in this section are based on Current Population Survey data that are not adjusted for income underreporting. A second complication is the difficulty of defining SPM-based poverty in a consistent way across the half century between 1967 and 2016. Recall that the SPM uses a poverty threshold based on the 33rd percentile of the distribution of core living expenses. Thus, the poverty threshold in the SPM is tied to changes in the standard of living of this low-income group. In contrast, the OPM poverty thresholds are adjusted over time only by rates of inflation.
Wimer et al. (2013) have estimated annual SPM thresholds going back in time to 1967, using available ASEC historical data. They have also constructed SPM thresholds that are anchored in current living standards and adjusted them backward in time only by inflation, as well as thresholds that are anchored in 1967 and then adjusted forward only by inflation. Though the SPM was designed to be a relative measure, whether to measure poverty in relative or absolute terms for purposes of historical analysis is an unsettled question. We use an anchored SPM (an absolute measure) here and in our analysis in Chapter 4 of the effects of changes in the labor market, family structure, and government programs on child poverty over time, because this measure allows us to abstract from changes in living standards. We anchor the measure in recent (2012) living standards to make it as comparable as possible with the TRIM3-SPM poverty estimates presented elsewhere in this report, which focuses on the current period.16Appendix D, 2-10 provides further discussion and illustration of child poverty trends using anchored and unanchored SPM measures.
Figure 2-11 shows both OPM- and anchored SPM-based child poverty rates from 1967 to 2016. As noted before, over this period OPM-based child poverty rates increased from 16.6 percent to 18.0 percent, while the anchored SPM indicates that child poverty actually decreased by nearly half—from 28.4 percent to 15.6 percent.17 SPM poverty rates are higher than OPM poverty rates in the earlier years of the period in part because of the higher SPM threshold and (to a lesser extent) because during that
16 These estimates were taken from a study (Wimer, 2017) commissioned by the committee for this report. Due to the relative nature of the SPM, historical changes in poverty could be at least partly due to changes in poverty thresholds (Wimer et al., 2013). Anchored measures of poverty apply current poverty thresholds to historic data by adjusting for inflation to isolate changes in family resources from changes in living standards. For more information, refer to Wimer et al. (2013).
17 As explained in Fox et al. (2015), an SPM poverty line anchored in 1967 living standards and subsequently adjusted for inflation annually yields estimates of poverty reduction that are similar to estimates anchored in current living standards and adjusted backward for inflation, like those reported in the figures and text.
period the tax system took more income from poor families with children than these families received from government as in-kind benefits. As we show in Chapter 4, much of the decline in SPM-based child poverty is due to increasingly generous government benefits. Because it does not count benefits from the Earned Income Tax Credit, SNAP, public housing, and housing vouchers, OPM-based child poverty rates include only cash transfers (like Supplemental Security Income [SSI] and the cash portion of TANF) and therefore fail to consider the largest portion of the social safety net. Consequently, trends in the OPM are not useful for drawing conclusions regarding changes in the well-being of children in the United States.
An alternative is to construct SPM poverty thresholds based on changes in living standards rather than inflation; this “historical SPM” also shows a substantial decrease in child poverty, but the decrease is only about half as large, or 25 percent (see Figure 2-15 in Appendix D, 2-10). The decrease in poverty is smaller because living standards at the 33rd percentile of the income distribution have increased over the last half-century by more than the cost of living. Figure 2-12 depicts historical trends in anchored SPM-based child poverty, near poverty, and deep poverty rates. As with the basic (under 100%) SPM poverty measure, shown in Figure 2-11, deep poverty rates had fallen by 2016 to nearly half of their 1967 levels. In the case of the line drawn at 150 percent of SPM, poverty rates fell by nearly 40 percent between 1967 and 2016. Strikingly, most of these three sets of declines occurred prior to the year 2000. It is also worth noting that SPM-based poverty rates declined for all three racial/ethnic groups: for Whites, Blacks, and Hispanics. (Historical trends in OPM- and SPM-based child poverty rates by race and ethnicity between 1970 and 2016 are presented in Appendix D, 2-8.)
CONCLUSION 2-5: When measured by the Official Poverty Measure, poverty rates changed very little between 1967 and 2016; by contrast, when measured by the anchored Supplemental Poverty Measure (SPM), they fell by nearly half over that period, due to the increases in government benefits. SPM-based rates of deep and near child poverty declined as well over the period, both overall and across subgroups of children defined by race and ethnicity.
Over the past several decades, researchers have developed the capacity to analyze child poverty across countries by using comparable microdata. The two most widely used sources of international data are the Luxembourg Income Study (LIS), which allows analysts to work with the microdata, and
the Organisation for Economic Co-operation and Development (OECD) poverty and income database, which is more up to date but provides only country-level statistics and relative poverty measures.
Early staff and committee discussions with the sponsors of this report revealed a particular interest in comparing child poverty rates across a subset of OECD English-speaking nations: Australia, Canada, Ireland, the United Kingdom, and the United States. These countries have income support systems that differ from those found in central and northern Europe, including Scandinavia (Esping-Anderson, 1990). Three of them are large and diverse nations (Australia, Canada, and the United States), while the other two (Ireland and the United Kingdom), though smaller in size, still exhibit some geographic and ethnic heterogeneity. We gauge
the comparative effectiveness of anti-poverty programs across these same countries in Chapter 4.
Most published international poverty comparisons use a poverty line defined by a given fraction of each country’s median income, such as 40, 50, or 60 percent.18 This is a relative poverty concept because it measures the fraction of families who have income that is low relative to overall income in the country. Families in a high-income, industrialized country might all have incomes that are higher than the incomes of families in a low-income country, but relative poverty could still be high in the former if the lower-income families there were “further away” from the country’s overall median income.19
OECD poverty statistics are typically based on a poverty line drawn at 50 percent of median income, a line we will call “OECD-50.” For this measure, household resources include money income and near-cash benefits minus taxes (including tax credits). Estimates of child poverty using the OECD-50 for the United States and the four English-speaking comparison countries are shown in the top bars of Figure 2-13 (labeled “Relative Poverty (OECD-50)”). Rates of child poverty using this relative measure are much higher in the United States than in these peer countries—more than twice as high as in Ireland and nearly 5 percentage points higher than in Canada, the country with the second-highest child poverty rates.
To explore the sensitivity of cross-national child poverty rates to the specific definition of child poverty, Figure 2-13 also shows poverty rates using two other measures. The first uses LIS data to set the poverty threshold for each country at the same percentile of the country’s income distribution as the SPM threshold in the U.S. income distribution. Since that point is at the 40th percentile of the income distribution, we label this measure “Relative Poverty (LIS-SPM-40).” Drawing the line at the 40th percentile lowers child poverty rates, but the country rankings are similar to those found with the OECD-50 measure of relative poverty.
The third measure is based on what is sometimes called “absolute” poverty. Absolute poverty measures the fraction of families in a country whose incomes fall below some fixed amount, regardless of how affluent the country is. For this reason, high-income countries will tend to have lower absolute poverty rates than lower-income countries. In our case, the dollar levels of the U.S. SPM poverty thresholds are translated into poverty thresholds in other countries using the purchasing power of the dollar relative
19 The SPM poverty measure is also relative, but it is based on the distribution of expenditures rather than income, and is set at a given (33rd) percentile of the expenditure distribution rather than at a fraction of the median.
to other countries’ currencies.20 Because the translations are based on purchasing power parity (PPP) data, we label this measure “Absolute poverty (LIS-SPM-PPP).” Appendix D, 2-11 discusses these measures in more detail.
As shown in the bottom panel of Figure 2-13, using an absolute poverty standard changes the country rankings somewhat. The United Kingdom now has the highest absolute poverty rate, followed by the United States, Ireland, Canada, and lastly Australia. The primary reason for this shift in rankings is that living standards are generally higher for U.S. children than for UK children, so a poverty line defined by U.S.-based income cuts the UK income distribution at a higher point than where it cuts the U.S. income distribution.
Finally, we compare rates of deep poverty and near poverty in the United States and these peer countries using the LIS and the absolute SPM poverty measure (see Figure 2-14). At 3.6 percent, the United States has by far the highest rate of deep child poverty, nearly twice the rate seen in the next-ranked nation (Australia, at 1.9%).21 By contrast, the United States is in the middle of the pack where near poverty is concerned (defining near poverty as 150 percent of the absolute SPM), with a rate of 29.2 percent. This near-poverty rate is considerably lower than what is seen in the United Kingdom (46.4%) and Ireland (37.2%), where the poverty line cuts their distributions at a much higher income level (see Appendix D, Figure D2-3), but it is higher than in countries with absolute living standards most similar to those in the United States: Australia (21.6%) and Canada (27.2%).
Poverty rates for children in single-parent families, in working families (except for the United Kingdom), and in immigrant families are higher in the United States than in the other comparison nations, even using the absolute LIS-SPM-PPP poverty rates. (These rates are shown in Figure 2-13 and Appendix D, 2-11.)
CONCLUSION 2-6: How child poverty rates in the United States rank relative to those in peer English-speaking developed countries depends on how poverty is defined. The United States has much higher rates of child poverty than these peer countries using relative, within-country measures of poverty. However, when an absolute poverty measure is used, child poverty rates in the United States are more similar to rates
20 The 2013 U.S. SPM translates into about $25,550 for two parents and two children. This amount is converted to other currencies using 2011 purchasing power parities (PPP) and national consumer price changes when years differ. The SPM poverty line income, on a household basis, ignoring health care costs and work expenses and other adjustments for COLAs and housing status, is about 40–41 percent of the U.S. median adjusted income on a comparable basis (Fox, 2017; Short, 2013; Wimer and Smeeding, 2017).
in peer countries. Rates of deep poverty, by contrast, are considerably higher for children in the United States than for children in these peer countries, whether absolute or relative measures are used.
Bollinger, C., Hokayem, C., Hirsch, B., and Ziliak, J. (2018). Trouble in the tails? What we know about earnings nonresponse thirty years after Lillard, Smith, and Welch. Journal of Political Economy, December. doi: 10.1086/701807.
Capps, R., Fix, M., and Zong, J. (2016). A Profile of U.S. Children with Unauthorized Immigrant Parents. Washington, DC: Migration Policy Institute.
Child Trends Data Bank. (2016). Educational Attainment: Indicators of Child and Youth Well-Being. Bethesda, MD: Child Trends. Available: https://www.childtrends.org/wp-content/uploads/2016/12/06_Educational_Attainment.pdf.
Council of Economic Advisors. (2018). Expanding Work Requirements in Cash Welfare Programs. Washington, DC: The White House. Available: https://www.whitehouse.gov/wp-content/uploads/2018/07/Expanding-Work-Requirements-in-Non-Cash-Welfare-Programs.pdf.
Eggebeen, D.J., and Lichter, D.T. (1991). Race, family structure, and changing poverty among American children. American Sociological Review, 56(6), 801–817.
Esping-Andersen, G. (1990). The Three Worlds of Welfare Capitalism. Hoboken, NJ: John Wiley & Sons.
Fox, L. (2017). The Supplemental Poverty Measure: 2016. Current Population Reports P60261. Available: https://www.census.gov/content/dam/Census/library/publications/2017/demo/p60-261.pdf.
Fox, L. (2018). The Supplemental Poverty Measure: 2017. Current Population Reports P60265. Available: https://www.census.gov/content/dam/Census/library/publications/2018/demo/p60-265.pdf.
Fox, L., Wimer, C., Garfinkel, I., Kaushal, N., and Waldfogel, J. (2015). Waging war on poverty: Poverty trends using a historical supplemental poverty measure. Journal of Policy Analysis and Management, 34(3), 567–592.
Furstenberg, F.F. (2014). Fifty years of family change: From consensus to complexity. The ANNALS of the American Academy of Political and Social Science, 654(1), 12–30. Available: https://doi.org/10.1177/0002716214524521.
Hardy, B., Smeeding, T., and Ziliak, J.P. (2018). The changing safety net for low-income parents and their children: Structural or cyclical changes in income support policy? Demography, 55(1), 189–221.
Hill, H.D. (2018). Chapter 6: Trends and divergences in childhood income dynamics, 1970–2010. In J.B. Benson (Ed.), Advances in Child Development and Behavior (vol. 54, pp. 179–213). Greenwich, CT: JAI Press.
Hokayem, C., Bollinger, C., and Ziliak, J. (2015). The role of CPS nonresponse in the measurement of poverty. Journal of the American Statistical Association 110 (511), 935–945.
Institute of Medicine and National Research Council. (2012). Medical Care Economic Risk: Measuring Financial Vulnerability from Spending on Medical Care. Washington, DC: The National Academies Press. https://doi.org/10.17226/13525.
Kennedy, S., and Bumpass, L. (2008). Cohabitation and children’s living arrangements: New estimates from the United States. Demographic Research, 19, 1663.
Korenman, S., Remler, D.K., and Hyson, R. (2017). Accounting for the Impact of Medicaid on Child Poverty. Background Paper for the Committee on Building an Agenda to Reduce the Number of Children in Poverty by Half in 10 Years, Board of Children, Youth and Families of the National Academies of Sciences, Engineering, and Medicine. Available: https://www.nap.edu/resource/25246/Korenman%20Remler%20and%20Hyson.pdf.
Lichter, D., Qian, Z., and Crowley, M. (2008). Poverty and economic polarization among children in racial minority and immigrant families. In D.R. Crane and T. Heaton (Eds.), Handbook of Families & Poverty. Thousand Oaks, CA: SAGE.
Manning, W., Brown, S., and Stykes, B. (2014). Family complexity among children in the United States. The ANNALS of the American Academy of Political and Social Science, 654(1), 48–65.
Martin, J.A., Hamilton, B.E., and Osterman, M.J. (2017). Births in the United States, 2016. Available: https://www.cdc.gov/nchs/products/databriefs/db287.htm.
Martin, J.A., Hamilton, B.E., Osterman, M.J., Driscoll, A.K., and Drake, P. (2018). Births: Final Data for 2016. Available: https://www.cdc.gov/nchs/products/databriefs/db287.htm.
Mather, M. (2010). U.S. Children in Single-Mother Families. Washington, DC: Population Reference Bureau.
Mathews, T.J., and Hamilton, B.E. (2016). Mean Age of Mothers Is on the Rise: United States, 2000-2014. Hyattsville, MD: National Center for Health Statistics.
McLanahan, S., and Jencks, C. (2015). Was Moynihan right? What happens to children of unmarried mothers? Education Next, 15(2).
Meyer, B.D., and Sullivan, J.X. (2017). Consumption and Income Inequality in the U.S. Since the 1960s. NBER Working Paper No. 23655. Cambridge, MA: National Bureau of Economic Research. Available: http://www.nber.org/papers/w23655.
Meyer, B.D., Mok, W.K.C., and Sullivan, J.X. (2009). The Under-Reporting of Transfers in Household Surveys: Its Nature and Consequences. NBER Working Paper No. 15181. Cambridge, MA: National Bureau of Economic Research. Available: http://www.nber.org/papers/w15181.
Migration Policy Institute. (2017). Children in U.S. Immigrant Families: Number and Share of the Total U.S. Child Population, by Age Group and State. Washington, DC: Migration Policy Institute. Available: https://www.migrationpolicy.org/programs/data-hub/charts/children-immigrant-families.
Moffitt, R.A., and Scholz, J.K. (2009, November). Trends in the Level and Distribution of Income Support. NBER Working Paper No. 15488. Cambridge, MA: National Bureau of Economic Research. Available: http://www.nber.org/papers/w15488.
National Academies of Sciences, Engineering, and Medicine. (2017). Promoting the Educational Success of Children and Youth Learning English: Promising Futures. Washington, DC: The National Academies Press. Available: https://doi.org/10.17226/24677.
National Center for Education Statistics. (2018). The Condition of Education: Characteristics of Children’s Families. Washington, DC: Institute of Education Sciences. Available: https://nces.ed.gov/programs/coe/indicator_cce.asp.
National Research Council. (1995). Measuring Poverty: A New Approach. Washington, DC: National Academy Press.
Pew Research Center. (2016). It’s Official: Minority Babies Are the Majority among the Nation’s Infants, but Only Just. Available: http://www.pewresearch.org/fact-tank/2016/06/23/its-official-minority-babies-are-the-majority-among-the-nations-infants-but-only-just.
Remler, D.K., Korenman, S.D., and Hyson, R.T. (2017). Estimating the effects of health insurance and other social programs on poverty under the Affordable Care Act. Health Affairs, 36(10), 1828–1837.
Renwick, T. (2015). Using the American Community Survey (ACS) to Implement a Supplemental Poverty Measure (SPM). Working Paper # 2015-09. Washington, DC: U.S. Census Bureau, Social, Economic and Housing Statistics Division.
Semega, J.L., Fontenot, K.R., and Kollar, M.A. (2017). Income and Poverty in the United States: 2016. Current Population Reports, P60-259. Available: https://www.census.gov/content/dam/Census/library/publications/2017/demo/P60-259.pdf.
Short, K. (2013). The Supplemental Poverty Measure: Examining the Incidence and Depth of Poverty in the U.S. Taking Account of Taxes and Transfers in 2012. Discussion paper. Washington, DC: U.S. Census Bureau.
Short, K., and Smeeding, T. (2012). Understanding Income-to-Threshold Ratios Using the Supplemental Poverty Measure: People with Moderate Income. SEHSD Working Paper No. 2012-18. Washington, DC: U.S. Census Bureau.
Summers, B.D., and Oellerich, D. (2013). The poverty-reducing effect of Medicaid. Journal of Health Economics, 32, 816–832.
U.S. Census Bureau. (2018). Annual Estimates of the Resident Population by Sex, Age, Race, and Hispanic Origin for the United States and States: April 1, 2010 to July 1, 2017. Washington, DC: U.S. Census Bureau.
Wheaton, L. (2008). Underreporting of Means-Tested Transfer Programs in the CPS and SIPP. Washington, DC: Urban Institute. Available: https://www.urban.org/research/publication/underreporting-means-tested-transfer-programs-cps-and-sipp.
Wimer, C. (2017). Child Poverty in the United States: Long-term Trends and the Role of Antipoverty Programs Using the Anchored Supplemental Poverty Measure. Commissioned by the Committee on Building an Agenda to Reduce the Number of Children in Poverty by Half in 10 Years. The National Academies of Sciences, Engineering, and Medicine, Washington, DC.
Wimer, C., and Smeeding, T. (2017). USA child poverty: The impact of the great recession. In Y.C.B. Cantillon, S. Handa, and B. Nolan (Eds.), Children of Austerity: The Impact of the Great Recession on Child Poverty in Rich Countries (pp. 297–331). Oxford: Oxford University Press.
Wimer, C., Fox, L., Garfinkel, I., Kaushal, N., and Waldfogel, J. (2013). Trends in Poverty with an Anchored Supplemental Poverty Measure. Columbia University Academic Commons. Available: https://doi.org/10.7916/D8RN3853.
Winship, S. (2016). Poverty after Welfare Reform. New York, NY: Manhattan Institute.
Wood, D. (2003). Effect of child and family poverty on child health in the United States. Pediatrics, 112 (Suppl. 3), 707–711.
Woods, T., and Hanson, D. (2016). Demographic Trends of Children of Immigrants. Washington, DC: The Urban Institute.
This page intentionally left blank.