Appendix D

Technical Appendixes to Select Chapters

APPENDIX D, 2-1
A BRIEF HISTORY OF POVERTY MEASUREMENTS IN THE UNITED STATES

The nation’s Official Poverty Measure (OPM), developed in the 1960s by Mollie Orshansky of the Social Security Administration, is based on research showing that, in the 1950s, the average family spent about one-third of its after-tax income on food (Fisher, 1992). Orshansky then multiplied the cost of the U.S. Department of Agriculture’s “basic” diet food plan by three to calculate poverty thresholds for families of different compositions and sizes (she used somewhat different methods for families of one and two people). Families with resources (regular money income as measured in a supplement to the Current Population Survey) below these amounts were considered to be poor. These thresholds have been adjusted for inflation every year using the Consumer Price Index. The official poverty rate is calculated in essentially the same way now as it was in the 1960s.

This approach to measuring poverty has numerous shortcomings: It is based on the now outdated assumption that families spend one-third of their post-tax income on food (today they spend less than one-half that amount); it fails to adjust for geographic differences in living costs; and, more importantly, it counts neither in-kind benefits nor refunded tax credits as income. Thus, the Earned Income Tax Credit (EITC), the Child Tax Credit, Supplemental Nutrition Assistance Program (SNAP) benefits, child care assistance, subsidized housing, and many other in-kind benefits are ignored in computing the poverty rate. It is also an absolute poverty

measure in that the poverty thresholds are updated for inflation but not for changes in the country’s standard of living.¹

These problems with the OPM were considered serious enough to generate congressional interest in an improved measure. Funding from a provision in the Family Support Act of 1988, which mandated a National Research Council study of a national minimum benefit standard for the now defunct Aid to Families with Dependent Children (AFDC) Program, combined with funding from the Bureau of Labor Statistics and the Department of Agriculture, supported the work of an NRC panel of poverty experts. This panel produced an influential report, Measuring Poverty—A New Approach, in 1995.

Despite the attention generated by that panel’s report, the OPM remains unchanged for many reasons, two of which are of major importance. First, numerous pieces of legislation stipulate that cash grants are to be provided to states, cities, and school districts, and the allocation of those grants is often based on the area’s poverty rate. Changing the poverty measure used in allocation formulas would affect the distribution of money among states and local areas, substantially in some cases.

Relatedly, the official poverty thresholds (actually a variant of them) are used to determine eligibility of families for a number of assistance programs. Second, defining a poverty level requires making judgments. Given the financial stakes of determining who is poor and who is not, Congress would certainly be mired in a long and contentious debate if it sought to set new poverty thresholds or change other aspects of the official measure.

Nonetheless, following the 1995 report, analysts at the Census Bureau and the Bureau of Labor Statistics began the complicated work of implementing many of the report’s recommendations for improving the measurement of poverty. By 2011, after more than a decade of study, the Census Bureau released a new method of defining poverty. The title of the new measure—the “Supplemental Poverty Measure” (SPM)—made it clear that the federal government, and in particular the Census Bureau, was not proposing to replace the OPM with a new one, but rather adding a new measure that would advance research and provide additional information for policy discussions.

___________________

¹ Because the inflation index used for the Official Poverty Measure (OPM) overstated price changes for a period, the official thresholds have actually been updated to some extent in real terms—that is, for changes in living standards. See Appendix D, 2-2, for a discussion of absolute versus relative poverty measures and the use of different inflation measures for adjusting thresholds.

APPENDIX D, 2-2
TYPES OF INCOME-BASED POVERTY MEASURES AND THE ADVANTAGES OF USING THE ADJUSTED SPM FOR POLICY ANALYSIS

Poverty measurement requires a set of decisions about the purpose and specifications of the measure. As background, Box D2-1 defines key terms in poverty measurement, such as threshold and resource concepts and whether the measure is intended to be absolute or relative. It also briefly defines types of poverty measures, including economic measures, such as those based on income or consumption, and other kinds of deprivation and hardship indexes.

The committee’s charge is to identify and estimate the benefits and costs of government policy and program options that can reduce child poverty and deep poverty in the United States within 10 years and to propose practical measures to do so given the available data. For this purpose, it is necessary to use an income-based economic poverty measure, which can clearly show the effects of one or another policy option on the adequacy of families’ resources. Other types of measures, such as deprivation and material hardship indexes, add to the picture of families’ living situations and well-being but present challenges to any straightforward estimation of the effects of government tax and transfer policies on them. This is also a limitation at present of consumption-based economic poverty measures—see Appendix D, 2-3.

The committee’s charge directs us to base our analyses on the income-based SPM and not on either the OPM or a consumption-based poverty measure. Unlike the Census Bureau, we use an adjusted SPM, in which we correct the underlying data for some types of income underreporting using the Urban Institute’s Transfer Income Model, Version 3 (TRIM3) microsimulation model. This appendix section describes and assesses the OPM, the SPM, and the adjusted SPM. It then discusses several contentious issues for income-based poverty measurement: relative versus absolute poverty; inflation adjustments; and the implications for income-based poverty, particularly deep poverty, of error in the underlying data source—the Current Population Survey Annual Social and Economic Supplement (CPS ASEC). See Appendix D, 2-3, for a discussion of consumption-based poverty measures, for which issues of relative versus absolute poverty, inflation adjustments, and data quality are also relevant.

Three Income-based Poverty Measures: OPM, SPM, Adjusted SPM

The OPM, the SPM, and the adjusted SPM used in this report are based on resources defined by a family’s income. A family is defined as poor if

family income is below a specific cutoff. The cutoff is chosen to indicate the income needed to attain a minimum or basic level of goods and services. This simple description masks the many specific decisions that are required to define an acceptable, feasible, and useful income-based poverty measure. Table D2-1 compares the OPM, the SPM, and the adjusted SPM on 11 dimensions: uses, measurement unit, threshold concept, threshold adjustments, threshold updating, treatment of health benefits and costs, resource measure, reference period, metric(s) estimated, data source, and data quality.

Three key differences between the OPM and the SPM merit fuller discussion: the type of measure (absolute or relative), the threshold concept, and the definition of family resources. Other important differences—in the adjustments to the thresholds for family composition and the unit of measurement—are briefly described in Table D2-1.

Type of measure. The OPM officially became an absolute measure in 1969, when proposals to update in real terms the original 1963 thresholds (based only on food needs—see next section) were turned down because policy makers were reluctant to show any increase in poverty.² Since then, the thresholds have been updated for inflation only, using the flagship series published by the Bureau of Labor Statistics—the All Items Consumer Price Index for All Urban Consumers, known as the CPI-U. Overestimation of inflation in the CPI-U, however, resulted in real increases to the thresholds in the 1970s and early 1980s (see discussion of inflation indexes below).

The SPM, in contrast, is a quasi-relative poverty measure in which the thresholds are based on needs for a specific set of goods—but a broader set than food—and are recalculated every year as spending on those goods changes. It is similar to an absolute measure because the thresholds are

___________________

² See https://www.census.gov/content/dam/Census/library/working-papers/1995/demo/fisher3.pdf. Between 1963 and 1969, changes in the costs of Economy Food Plan were used to update the thresholds.

TABLE D2-1 Dimensions of Three Income-Based Economic Poverty Measures: Official Poverty Measure (OPM), Supplemental Poverty Measure (SPM), Adjusted SPM

NOTE: ACS = American Community Survey; CPI-U = Consumer Price Index-Urban Consumers; NRC = National Research Council; SIPP = Survey of Income and Program Participation; SNAP = Supplemental Nutrition Assistance Program; SSI = Supplemental Security Income; TANF = Temporary Assistance to Needy Families.

^a See, e.g., https://www.urban.org/sites/default/files/publication/24401/904570-Expanded-Poverty-Measurement-at-the-State-and-Local-Level.PDF [January 2019].

SOURCE: Compiled by committee staff from Census Bureau and other sources.

based on a specific set of needs, but it has an element of a relative measure because the thresholds change as spending on those goods changes over time in the U.S. population. The SPM consequently does not require an inflation index per se except to make all 5 years of Consumer Expenditure Survey (CE) data used in constructing the thresholds consistent in real terms.³ The recalculation of SPM thresholds every year is intended to produce a conservative or quasi-relative measure compared with a relative measure, such as a percentage of median income (see discussion of absolute vs. relative measures and inflation indexes below).

Threshold concept. The OPM thresholds, developed originally for 1963, represent a type of “expert budget.” Specifically, U.S. Department of Agriculture nutritionists costed out a basic food plan (the Economy Food Plan) that would be minimally nutritious and palatable. Mollie Orshansky at the Social Security Administration multiplied the cost of the food plan for different sizes and compositions of three-or-more-person families by three to allow for all other needed spending, using the spending patterns of the average family in 1955 as the basis for the multiplier. She used a different method for two-person families and single individuals, with lower thresholds when such a family was headed by someone aged 65 or older, on the grounds that their food needs were less.

The SPM thresholds, in contrast, allow for the costs of a much broader bundle of basic needs, including food, clothing, shelter, and utilities (FCSU), plus a small multiplier (1.2) to account for other necessary items, such as non-work-related transportation and personal care. The FCSU component is based on actual consumer expenditures for families with any number of adults and two children at the 33rd percentile of the distribution. A lower percentile is not used because families’ spending could be constrained by their income, and therefore a lower percentile could underestimate their basic needs. The thresholds for two-child families are calculated separately for renters, homeowners with a mortgage, and homeowners who own their homes free and clear. Homeowners without mortgage debt have a lower threshold than the other two groups because their out-of-pocket housing needs are less. A formal equivalence scale is used to adjust the SPM thresholds for different sizes and composition of families (see Appendix D, section 2-4), and, finally, the thresholds are adjusted for geographic variations in the cost of housing (see Appendix D, section 2-5).

Resource concept. The OPM has a limited measure of family resources, namely, regular money before-tax cash income, which was the definition used for data collection in the CPS. The definition was not unreasonable at the time before the expansion of tax credits and in-kind benefit programs.

___________________

³ That is, the thresholds rise over time in response to price increases only to the extent that the prices of the basic goods rise.

The SPM defines resources (see Appendix D, section 2-6) to recognize that substantial government assistance is provided to families through the tax system and in-kind transfers, such as housing subsidies, free school lunches, and SNAP (formerly food stamps). The SPM uses a disposable money and near-money income resource definition, not only including the cash value of such programs as SNAP and lump sums received as tax credits, but also recognizing that some expenses are nondiscretionary. Specifically, the SPM definition subtracts work-related expenses including transportation and child care, child support payments to another family, out-of-pocket medical care payments (including premiums, co-pays, deductibles, and uncovered care), and net taxes (federal income and payroll and state income) after credits. This definition provides a much more realistic picture of families’ resources for their everyday needs for food, clothing, shelter, utilities, and a little more.

The adjusted SPM is the same as the SPM published by the Census Bureau with one important exception. The TRIM3 model used by the committee for its analyses adjusts three types of government transfers for underreporting by survey respondents, using aggregate totals from administrative records as benchmarks. These three sources are SNAP, Supplemental Security Income (SSI), and Temporary Assistance to Needy Families (TANF). They contribute importantly to low-income families’ resources and also suffer from significant and increasing underreporting in the CPS ASEC and other surveys (see discussion below).

Key Issues for Economic Poverty Measurement

This section discusses three key issues that are sources of debate in evaluating not only the OPM, the SPM, and the adjusted SPM, but also consumption-based measures (see Appendix D, section 2-3). They are absolute vs. relative measures; inflation indexes; and the implications of the well-documented underreporting of income in surveys for poverty measurement, particularly deep poverty.

Absolute vs. Relative Measures

The OPM is intended to be an absolute poverty measure even though the thresholds were inadvertently adjusted in real terms because of problems with the CPI-U used to update them for inflation. Recent work on consumption-based poverty also uses an absolute measure, or more precisely, an anchored measure (see Appendix D, section 2-3).

An absolute measure may make sense for monitoring poverty trends for a period of time because it affords a fixed standard of need. In contrast, relative measures offer a moving target, which may be less intuitive to the public and policy makers. Maintaining an absolute measure over long

periods, however, can give results that do not square with contemporary perceptions of deprivation and are not helpful for policy. The National Research Council (1995, Figure 1-1) illustrated this phenomenon by comparing the OPM two-adult/two-child threshold with a subjective threshold derived from public opinion and a threshold (similar to that used in many countries) of 50 percent of median after-tax money income, all in constant 1992 dollars. In 1963, all three thresholds were in agreement, but in 1947, the other two thresholds were only 68-75 percent of the OPM threshold. In contrast, by 1970, the other two thresholds exceeded the OPM threshold by 20 percent (National Research Council, 1995, Tables 2-3, 2-4).

At the heart of the difference between absolute and relative poverty measures is whether the basic needs that society deems that every family should have should be allowed to change over time. This is most clearly illustrated by the concept of deprivation and material hardship (see Appendix D, Box 2-1). Arguments have been made, by looking at such deprivation measures as material hardship, that the poverty rate is not as high as the OPM or even an income-based measure corrected for income underreporting would indicate (see, e.g., Meyer and Sullivan, 2011). For example, more lower-income families had air conditioning and other appliances in 2009 compared with 1980 (the period studied in Meyer and Sullivan, 2011), and using an absolute measure poverty declined between those years. More generally, living standards overall have increased over the past 30 years for the entire population. Yet that increase does not mean that an absolute poverty measure, established 30 or 40 years in the past, is necessarily preferable to a relative or quasi-relative measure, because what is regarded as a basic need by society generally increases along with living standards.

An historical example is telling in this regard: In 1940, 45 percent of U.S. households lacked complete plumbing facilities compared with only 7 percent in 1970 shortly after the OPM was adopted.⁴ Yet it is unlikely that a 1940 poverty threshold would have been set so as to result in a 45 percent poverty rate. Conversely, the OPM threshold set in 1963, which was universally felt to be “right” at the time (see National Research Council, 1995, p. 110), gave a poverty rate for 1970 of 12.6 percent, almost twice as high as the percent of households then with inadequate plumbing. Moreover, as noted above, by 1969 the OPM threshold itself was viewed as too low, and today basic budgets constructed by nongovernmental organizations, when expressed in comparable terms, are as high as the SPM thresholds, which represent a real increase over the OPM thresholds (see, e.g., the “Household

___________________

⁴ See https://www.census.gov/hhes/www/housing/census/historic/plumbing.html.

Survival Budget” calculated for counties in 18 states as part of the United Way ALICE Project).⁵

The National Research Council (1995) study that recommended a new poverty measure, which with some modifications became the SPM (see Appendix D, 2-1), did not recommend a completely relative measure, such as a percentage of median income, as used in many other countries (see Appendix D, 2-11). Instead, it recommended what it termed a “quasi-relative” updating procedure, based on changes in consumption of basic necessities (food, clothing, shelter, and utilities) in the lower part of the distribution of consumer expenditures as measured in the CE. The 1995 study (Ch. 2) cited evidence that subjective thresholds, based on public opinion, lagged increases in median income as support for its supposition that a quasi-relative updating procedure would be more acceptable to the public and policy makers. It further cited evidence that real increases in consumption on necessities lagged real increases in total expenditures as measured in the Bureau of Economic Analysis Personal Consumption Expenditures series as justification for its recommended updating procedure.

Inflation Indexes

All poverty measures that have any absolute element, including anchored measures, require a measure of price change or inflation to keep their thresholds constant in real terms. It is well known that the CPI-U, which is used to adjust the OPM thresholds each year, has overstated inflation in the past. BLS maintains the CPI-U-Research Series (CPI-U-RS) to provide a historical series that estimates improvements in the CPI-U back through 1978, with the CPI-U-RS reestimated annually to incorporate as many additional improvements in the CPI-U as is feasible. Most, but not all, of the improvements reflected in the CPI-U-RS produced lower inflation rates relative to the corresponding CPI-U, often significantly so prior to 2000. Since 2001, the two series are very similar.

In 2002, BLS introduced a chained CPI series, the C-CPI-U, which lowered inflation even more than the CPI-U-RS by more fully capturing consumers’ ability to substitute among items in the face of price changes. In 1999, a method for capturing substitution within item categories was introduced into the CPI and CPI-U-RS; the C-CPI-U, also beginning in

___________________

⁵ ALICE stands for Asset-Limited, Income-Constrained, Employed, as the focus of the ALICE Project is on the working poor; see https://www.unitedwayalice.org/overview. Note that alternative basic budgets typically need to have several components subtracted for comparability with the SPM thresholds—e.g., child care, work-related transportation, medical care, and taxes must be subtracted from the ALICE Household Survival Budget because these items are subtracted from SPM resources and are therefore not included in SPM thresholds. Interestingly, the Household Survival Budget includes smartphone costs as a basic need.

1999, uses in addition a method for capturing substitution among item categories.⁶ Between 1999 and August 2017 (the latest estimate available), the C-CPI-U has averaged 1.9 percent inflation per year, compared with 2.2 percent for the CPI-U and the CPI-U-RS between 1999 and 2017.⁷

Whether the official poverty thresholds should be adjusted with the C-CPI-U rather than the CPI-U, or whether they should be adjusted with an index that exhibits even less inflation than the C-CPI-U, are questions for further research.⁸ The SPM does not require an inflation index, except to make all 5 years of CE data used in the threshold calculations expressed in the same dollars and except when SPM thresholds are anchored for purposes of historical analysis (see Appendix D, 2-10), although its thresholds do rise over time as the price of its basic necessities rise in addition to rising because of increased consumption (in real terms) of these necessities.

Data Quality and Deep Poverty

Chapter 2 notes the extent of income underreporting in the CPS ASEC for many types of income and not just those accounted for in the adjusted SPM. Researchers (e.g., Winship, 2016; Appendix D, 3-1) have cited underreporting as a reason to doubt estimates of deep poverty, even with TRIM3 adjustments to some types of income. Meyer and Sullivan (2012a; online appendix table 10), using the 2010 CE, find that families in deep poverty with an SPM measure (unadjusted for underreporting) do not lack for amenities (e.g., major appliances) any more than officially deeply poor families. Further analysis would be required to assess these findings using the adjusted SPM measure, which in our analysis yields a very low deep poverty rate for children in 2015—2.9 percent—compared with a rate of 4.9 percent with the unadjusted SPM and a rate of 8.9 percent for the official measure.

Yet, cognizant that problems in income reporting could affect our estimates of adjusted SPM deep poverty, given the low thresholds involved (about $13,000 for two-adult/two-child renters and owners with a mortgage in 2015), we investigated further the characteristics of deeply poor

___________________

⁶ See https://www.bls.gov/cpi/additional-resources/chained-cpi-questions-and-answers.htm.

⁷ The C-CPI-U historical series is available at: https://www.bls.gov/cpi/additional-resources/chained-cpi-table24C.pdf; the CPI-U and CPI-U-RS calculations are derived from a spreadsheet provided by Bruce Meyer and James Sullivan to the committee in an e-mail, January 8, 2019.

⁸ Indeed, there are many reasons for research into the effects of using different price indices in measuring well-being over time for specific populations. For instance, some think that the prices paid by poor people differ from those paid by rich people.

families with children in 2015.⁹ We found evidence of anomalies resulting from the treatment of self-employment income in the CPS ASEC. Thus, deeply poor children in families with self-employment income made up 12.6 percent of the deeply poor group (15% of these families, or 1.9% of all deeply poor, had negative self-employment income). The CPS ASEC asks respondents for their profit or loss from self-employment, which means that many self-employed people may report accounting losses (or a lesser amount of profit) when they have not, in fact, experienced real declines in their ability to meet their needs.

We also found anomalies in the case of interest and dividend income. Thus, 28.1 percent of deeply poor children lived in families receiving dividends or interest (although two-thirds of these families had their dividend or interest income imputed by the Census Bureau). Examining the distribution of interest and dividends (imputed and reported), the dollar amounts for most families were small (the 75th percentile value for deeply poor children was $142, compared with $109 for poor children and $1,200 for all children). But there were outliers with relatively high amounts of dividend income (the 90th percentile value for deeply poor children was $1,559, compared with $1,001 for poor children and $5,879 for all children). These results suggest that some imputations of high values for interest and dividend income may have been made erroneously, especially given that the 90th percentile values for deeply poor children in families with reported interest and dividends were several hundred dollars lower than those for imputed and reported combined. It is also likely that at least some deeply poor families with income from such sources as self-employment and interest and dividends were not usually poor, let alone deeply poor, but had a below-average year with high out-of-pocket medical care expenditures or work expenses that pushed them into deep poverty.

Our very preliminary analysis suggests that further research is needed into the characteristics of children living in families classified as deeply poor under the adjusted SPM. Research is also needed on better ways to collect self-employment income in the CPS ASEC and to evaluate imputation procedures to be sure they take relevant variables into account and do not impute high values of such sources as interest and dividends inappropriately.

Advantages of the Adjusted SPM for Policy Analysis

From what is known about the strengths and weaknesses of the three income-based poverty measures reviewed above, the committee concludes

___________________

⁹ Results presented for self-employment and interest and dividend income were performed at the committee’s request by the Urban Institute using TRIM3.

that for its purpose of analyzing government policies and programs that can reduce income-based child poverty:

• The SPM is preferable to the OPM because, among other improvements, it includes “near-cash” benefits from SNAP, housing subsidies, and other assistance programs and from refundable tax credits. It also accounts for costs of work, such as day care. By contrast, the OPM counts only regular gross money income and thereby overstates the extent of child poverty and underestimates the positive effects of government programs in reducing child poverty. Critical for the committee’s purposes is that the OPM cannot be used to estimate the effects on child poverty of policy or program options that involve in-kind benefit programs or tax credits.
• The adjusted SPM is preferable to the unadjusted SPM because of its corrections for underreporting of income from SNAP, SSI, and TANF in the CPS ASEC, which has worsened in recent years.
• The adjusted SPM shares some of the drawbacks of the SPM (and the OPM) as a measure of income, namely:
  • The adjusted SPM (also the SPM and OPM) underestimates the poverty-reducing effects of medical care coverage, such as from Medicaid and Medicare (see Ch. 7 for a proposed solution).
  • The adjusted SPM does not correct for underreporting of sources of income other than SNAP, SSI, and TANF, including underreporting of market income.
  • The adjusted SPM (also the SPM and OPM) may overestimate the extent of deep poverty as a result of how certain kinds of income are collected in the CPS ASEC, such as self-employment losses, and as a result of error in imputations for families reporting receipt but not amounts of some income types. These problems could be ameliorated with research and improvements to the quality of CPS ASEC income data (see Ch. 9).

Finally, we note that parts of our report rely on the OPM or an anchored (unadjusted) SPM. For example, our review in Chapter 3 of the literature on the consequences of child poverty for outcomes necessarily uses the OPM, because the SPM has not been available for a long enough period for outcomes research. While the SPM (even better, a fully adjusted SPM) would be an improvement over the OPM, any reasonable measure of low income can suffice for this type of research. Our review of trends in child poverty in Chapter 4 uses an unadjusted SPM anchored in 2012, because TRIM3 adjustments to the SPM are not available historically. Consequently, the poverty rates shown are somewhat higher than they

would be with the adjusted SPM, but there is no reason to believe that the overall trends are invalid (see Appendix D, 2-10). None of our simulated packages in Chapters 5 and 6—and therefore none of our calculations for which programs would reduce poverty by 50 percent—use the OPM or an anchored SPM. In addition, as experience is gained with the SPM going forward, particularly if our recommendations in Chapter 9 for improvements to the CPS ASEC are adopted so that the SPM can be derived from complete income information, we are confident that the SPM will continue to be useful and informative for research and policy.

APPENDIX D, 2-3
CONSUMPTION-BASED POVERTY MEASURES

All of the economic poverty measures discussed in Appendix D, 2-2, use variants of an income-based measure of resources. An alternative approach to economic poverty measurement is to use a family’s consumption rather than its income to capture family resources. In this appendix, we discuss the definition of consumption poverty, how it has been measured, and the arguments in favor of a consumption-based poverty approach. We also discuss a number of problems with implementing a measure of consumption-based poverty using data currently available from the federal statistical system.

Most economists believe that consumption is a better measure of well-being than income because their theories consider a family’s well-being (“utility”) to be generated by the goods and services consumed by the family. If, over the course of every month, a family consumes exactly 100 percent of its monthly income, then income and consumption would be equal and would indicate the same level of well-being. But incomes can fluctuate from period to period. Provided that a family is able to save its income and/or access credit from one period to the next, it should be able to “smooth” consumption against income fluctuations, which would produce more stable and consistent amounts of monthly consumption than would be indicated by monthly income. If smoothing is feasible for families, then consumption should provide a better measure of well-being.

In practice, however, low-income families have little in the way of assets and savings (see Ch. 8), so it is unclear whether the low-income families with children who are the focus of our report can do much, if any, smoothing (Hurst, 2012, p. 191). Indeed, to the extent that families facing declining income maintain their consumption by such strategies as unsecured credit, pay-day loans with high interest rates, and the like, a consumption-based poverty measure may not provide as timely an indicator of when low-income families are under increasing financial stress as an income-based poverty measure, assuming good measurement of income.

As detailed in the main body of Chapter 2, a practical challenge with income poverty measurement in the United States is significant underreporting of government transfers and other kinds of income in the CPS ASEC and other household surveys (Meyer, Mok, and Sullivan, 2009, 2015; Moffitt and Scholz, 2009; Wheaton, 2008). A potential advantage of a consumption-based definition of poverty stems from more complete survey reports of expenditures than income. Thus, Meyer and Sullivan (2003, 2011) find that expenditures (and consequently consumption, which is derived from expenditures) appear to be better measured than income in the CE for lower-income Americans (although CE income is less well reported than CPS ASEC income).¹⁰ It is not known how much of the difference is a result of underreporting of conventional sources of income in the CE versus families’ not being asked to report or not reporting less conventional sources, such as unsecured credit and gifts or short-term loans from relatives and friends, as income.¹¹

The CE has a number of drawbacks for measuring consumption poverty. It collects data on expenditures and asset holdings but not consumption per se. A comprehensive consumption measure requires imputing service flows to such assets as housing, vehicles, and consumer durables (e.g., appliances) and may also involve subtracting some types of expenditures because they are viewed as investments or for other reasons (see below). As with income data in all government household surveys, expenditures in the CE are underreported and are subject to important measurement error, attrition bias, and nonresponse bias (National Research Council, 2013).

The CE also has much smaller sample sizes than income surveys such as the CPS ASEC and especially the American Community Survey (ACS).¹² Thus, in its current form, the CE is not well suited to generate subnational estimates for poverty; in fact, the public-use version of the CE does not even identify state of residence. The CE data, in their current form, also are challenging to work with because they must be assembled for five quarters to measure consumption during a calendar year for a consumer unit (a family

___________________

¹⁰ Once the CE began in 2004 to impute amounts for people who said they had income but did not provide an amount, the ratio of CE income to CPS/ASEC income rose from under 80 percent to about 95 percent across all income types; see https://www.bls.gov/cex/twoyear/200607/csxcps.pdf, text Table 4.

¹¹ The CE Interview Survey questionnaire module on income asks about lump sum payments from “persons outside your household” as part of a broad question on lump sum income; the CE Interview Survey asset and liability module asks about balances on credit cards, student loans, and all other loans, including personal loans, but this information is not integrated with “income” for purposes of comparing with expenditures (see https://www.bls.gov/cex/capi/2017/2017-CEQ-CAPI-instrument-specifications.pdf).

¹² The CE Interview Survey obtains about 7,500 consumer unit interviews per quarter, compared with the CPS ASEC’s 94,000 household interviews per year and the ACS’s 2.2 million household interviews per year.

or one or more people in a household who share income). Alternatively, quarters of expenditures can be pooled and annualized. This approach (used by Meyer and Sullivan in their studies—see below) maximizes the available sample but may not produce the same results as if expenditures were constructed for the same consumer units over the year. The Bureau of Labor Statistics (BLS) has a program under way to redesign the CE to improve its measurement of expenditures and related information, but implementation will take a number of years, and there is unlikely to be expansion of the sample.¹³

Most recent research on using the CE to measure trends in consumption inequality that attempts to correct for the CE’s measurement errors shows that consumption inequality tracks income inequality very closely through the mid-2000s (Attanasio and Pistaferri, 2016). Meyer and Sullivan (2017) show that the two measures diverge after approximately 2006, although they use a nonstandard price index (see below), which accounts for a good part of the divergence.

Meyer-Sullivan Consumption-Based Poverty Measure

Meyer and Sullivan, in a series of studies (Meyer and Sullivan, 2012b, 2017, 2018), use the CE to measure consumption poverty across groups and over time. They construct poverty thresholds by finding the threshold (after equivalizing consumption using equivalence scales from National Research Council, 1995) that leads to the same consumption and income poverty rates in some base year. They adjust the threshold from one year to the next using an inflation index that subtracts 0.8 percentage points per year from the All Items Consumer Price Index for All Urban Consumers-Research Series (known as the CPI-U-RS—see Appendix D, 2-2). Meyer and Sullivan (2012b) term this a “bias-corrected” CPI-U-RS (see further discussion below).

Using 1980 as the base year, Meyer and Sullivan (2018, Table 3) find that in 2017, 3.5 percent of children are poor based on their measure of consumption poverty compared with 9.4 percent based on their measure of income poverty. They measure income poverty using an after-tax-and-transfer resource measure that has some similarities to but is not the same as the SPM (see Box D2-2). Comparing the two series over time, their consumption and income poverty measures track each other fairly closely until 2000, when their child income poverty measure flattens out (as does the SPM), whereas the consumption-based measure of child poverty continues to decline steadily. They use the same inflation index, namely, their bias-corrected CPI-U-RS, for both series, so the sources of the difference

___________________

¹³ See https://www.bls.gov/cex/ce_gemini_redesign.pdf.

in trends must be due to other factors. Some possible explanations include these: the CE could have experienced an increasing rate of income underreporting after 2000 (similar to the CPS ASEC);¹⁴ homeowners could have given increasingly inflated estimates of the rent they expect their homes would bring;¹⁵ other expenditures could have become better reported over time;¹⁶ consumption, perhaps supported by credit, could have grown markedly faster than income; or some combination of these and other factors.

Comparing Meyer and Sullivan’s consumption measure for the whole population with the OPM using CE income data (they do not show results for the OPM comparison separately for children), the two measures track each other fairly closely until 1990, when the OPM poverty rate remains fairly steady, whereas the consumption rate declines steadily and substantially. By 2017, OPM poverty for the entire population was 12.3 percent compared with 7.0 percent for their income measure compared with only 2.8 percent for their consumption measure (Meyer and Sullivan, 2018, Table 1). The continued higher poverty rate for the OPM not only reflects the factors that account for the difference between Meyer and Sullivan’s income and consumption measures, but also two other factors: (1) increases in noncash benefits (e.g., SNAP) and tax credits over the period, which are not included in the OPM resource measure; and (2) the “bias-corrected” CPI-U-RS price index used by Meyer and Sullivan in contrast to the OPM, which uses the CPI-U (see further discussion below).

Meyer and Sullivan’s work uses an anchored measure of consumption-based economic poverty. In anchored measures, the starting (or ending) threshold value is selected to facilitate analysis of trends rather than on the threshold’s merits in level terms (see Appendix 2-2). For their series anchored in 1980, Meyer and Sullivan selected their threshold to give the same 13 percent poverty rate as the OPM. Consequently, their threshold was at the 13th percentile of the distribution of their measure of consumption, which amounted to $8,100 for a two-adult/two-child family in 1980

___________________

¹⁴ Meyer, Mok, and Sullivan (2009, Tables 2-9) show increased underreporting in both surveys over the 1990s and through 2006–2007 compared with earlier years for transfer program income, including SNAP and Temporary Assistance to Needy Families. Whether 2000 is the precise inflection point for marked deterioration in reporting would require closer examination of the data.

¹⁵ With regard to housing, the ratio of the CE estimate of imputed homeowners’ rent to the comparable estimate from the Bureau of Economic Analysis (BEA) personal consumption expenditures (PCE) series rose continuously from the mid-1990s to as high as 120 percent in the mid-2000s, falling back during the Great Recession (see Bee, Meyer, and Sullivan, 2012, Figure 1a). The ratio has stayed constant for 2014–2018 at 110 percent (see https://www.bls.gov/cex/cepceconcordance.htm).

¹⁶ The evidence on the accuracy of CE reporting of various expenditures, relative to the BEA PCE series, beginning in 2000 is mixed—with some types of expenditures exhibiting better reporting and others worse (see Bee, Meyer, and Sullivan, 2012, Figures 1a–1i).

dollars. That threshold was 97 percent of the official poverty threshold of $8,350 for 1980.¹⁷ The Meyer and Sullivan approach thus shares the OPM defect of using absolute needs in a particular year and then deriving poverty rates in other years without any direct assessment of whether needs are changing, unlike what is done in the SPM.

Further, Meyer and Sullivan did not assess basic consumption needs against their 1980 thresholds to see if the thresholds made sense relative to living standards at the time. Yet by 1980 it was clear that the OPM

___________________

¹⁷ The 97 percent figure is from Meyer and Sullivan (2012b, footnote 7); 1980 official poverty thresholds are available at https://www2.census.gov/library/publications/1982/demographics/p60-133.pdf; the threshold cited in the text is for “other” nonfarm families with four members, including two children.

thresholds fell considerably below other kinds of thresholds, such as those based on one-half of median income or those derived by asking samples of people their assessment of a poverty line (so-called subjective thresholds—see Appendix 2-2).

As it turned out, using their “bias-corrected” CPI-U-RS inflation factors, Meyer and Sullivan’s two-adult/two-child threshold for the current end point in their series (2017) was $17,765, which was only 71 percent of the comparable OPM threshold of $24,858.¹⁸ Yet they intend their

___________________

¹⁸ The Meyer and Sullivan threshold for 2017 was derived using deflators provided by Bruce Meyer and James Sullivan in an e-mail communication, January 9, 2019; 2017 official poverty thresholds are available at https://www.census.gov/content/dam/Census/library/publications/2018/demo/p60-263.pdf.

series not only for analytic purposes (as in the committee’s use of anchored SPM thresholds—see Ch. 2 and Appendix 2-10), but also as the basis for a substantively meaningful poverty series for policy makers. Given how their anchored thresholds were developed, it is hard to know how to assess the face validity of their thresholds either in 1980 or today.

Meyer and Sullivan’s use of their “bias corrected” CPI-U-RS is a major reason for the marked decline in poverty observed in their consumption-based measure. It is well known that the CPI-U has overstated inflation in the past and that BLS has endeavored to correct the CPI-U going forward. BLS produces the CPI-U-RS as a historically comparable series back to 1978 that incorporates the latest CPI-U improvements. BLS also, beginning in 1999, has produced a chained CPI-U (C-CPI-U), which corrects for a remaining source of overstatement of inflation in the CPI-U (the C-CPI-U averages about 0.3 percentage points per year below the CPI-U—see Appendix 2-2).

Meyer and Sullivan rely on several studies of bias in the CPI-U for their decision to subtract 0.8 percentage points per year from each year’s growth in the CPI-U-RS. The studies include Berndt (2006), the Boskin Report (Advisory Commission to Study the Consumer Price Index, 1996), and Hausman (2003). Berndt (2006), which reviewed what was done to improve the CPI-U following the Boskin report and other studies, found that BLS had made and was continuing to make major improvements. Meyer and Sullivan’s bias correction, which is applied at the same rate every year, does not have a direct basis in any particular prior study.

Whether there is justification for the sizeable correction Meyer and Sullivan make for upward bias in the CPI-U and CPI-U-RS is unclear and would require more study.¹⁹ More study would also be needed to investigate the optimum schedule for updating poverty thresholds in real terms—continuously as in the SPM and percentage-of-median income measures (which do not rely on an inflation measure); close to 40 years (as in the Meyer-Sullivan consumption measure); 55 years as would be the case for the OPM if the CPI-U had not been found to overestimate inflation; or a shorter interval. Suffice it to say that the use of the bias-corrected CPI-U-RS by Meyer and Sullivan to keep their 1980 threshold constant over almost 40 years produces contemporary thresholds and poverty rates that seem unrealistically low compared with other thresholds and rates.

It should also be emphasized that the method of anchoring used by Meyer and Sullivan makes the actual poverty rate in any given year an

___________________

¹⁹ Researchers often prefer the PCE deflator produced by BEA, which generally increases at a lower rate than the CPI-U (see, e.g., Winship, 2016). It also differs significantly from the CPI in scope (e.g., including medical care costs paid for by insurers) and other features, so that its advantages for adjusting poverty thresholds are not clear. See, for example, Johnson (2017).

arbitrary function of the anchoring year chosen. For example, if Meyer and Sullivan anchored their consumption poverty series in 2015 instead of 1980, they would conclude that the poverty rate in 2015 is the same as the OPM poverty rate—13.5 percent—and not the 3.4 percent they obtain for 2015 when anchoring in 1980. But that would imply that their poverty rate in 1960–1961 was as high as 62 percent (see Meyer and Sullivan, 2018, Table 2). Thus, their inflation index results in either implausibly high 1960–1961 poverty rates or implausibly low contemporary poverty rates, depending on the anchoring year chosen.

Reasons to Use the Adjusted SPM for This Study

For the purposes of our report, we do not use a consumption measure of poverty for four reasons:

1. The TRIM3 model adjusts the CPS ASEC for underreporting of three major types of transfer income, thereby addressing a large component of the problem with the use of income survey data.
2. The available data on consumption have the measurement issues discussed above.
3. Because the congressional charge to our committee is to assess how current and alternative transfer and other programs might change poverty, using a consumption poverty measure would require knowing how changes in those programs would affect consumer expenditures, and the research base on that relationship is scant and far from sufficient to use as a basis for simulations. Moreover, while the effect of transfers on income is conceptually straightforward and mostly mechanical, the effect of transfers on consumer spending requires understanding individual behavior and necessarily requires a higher level of research understanding than for income.
4. The Statement of Task for our committee directs us to use a specific income-based measure of poverty.

Taking this all into account, income poverty measured with the adjusted SPM is the appropriate measure for our use. The ability to incorporate corrections for underreporting of government transfers, using the TRIM3 model, is clearly crucial. A longer-term solution is to invest in improving our household surveys for income measurement—see Chapter 9. In fact, there is significant effort at the Census Bureau to incorporate administrative data into household surveys to do just that. Initial work by Meyer and coauthors (Meyer and Mittag, 2015; Meyer and Wu, 2018) shows that this is feasible and can lead to important findings on the measurement of

poverty and the evaluation of the antipoverty effects of government tax and transfer programs.

Work to improve the CE for measurement of expenditures and related information, as is currently under way at BLS, is also a worthwhile investment in an important component of the suite of essential federal statistics for research, public understanding, and policy analysis. The 1995 National Research Council report that recommended what became the SPM in fact recommended (p. 13) improvements in the CE that could support consumption-based poverty measures. Further research on how to relate changes in government policies and programs to consumption would also be very worthwhile.

APPENDIX D, 2-4
HOW EQUIVALENCE SCALES ARE USED TO ADJUST POVERTY THRESHOLDS

The resources needed for a family to achieve a basic standard of living, however defined, vary with the size and composition of the family. Expenditures on some basic need categories, such as shelter and utilities, may not increase dramatically with a marginal increase in household size; meanwhile, expenditures on other basics, such as food and clothing, may be more sensitive to the number of people in the household.²⁰

Since the amount spent on some necessities by larger families is greater than the amount spent by smaller families, poverty thresholds based on a 5-year moving average for consumption of food, clothing, and shelter must be adjusted to reflect the differences. Equivalence scales are typically used to make these adjustments so that families of various sizes and composition may be compared on as equal a basis as possible. The OPM thresholds, however, were constructed using a different approach (see National Research Council, 1995, Ch. 3).

As described in the first section of Chapter 2, for the SPM the poverty threshold is based on the 33rd percentile of expenditures on FCSU for resource units (families),²¹ multiplied by 1.2.²² The SPM poverty threshold is estimated using 5 years of data from the CE on out-of-pocket FCSU spending by household units with one or more adults and exactly two children; this is referred to as the resource threshold for the reference family.

___________________

²⁰ Although, even in these cases, economies of scale may be realized. Consider hand-me-downs, which allow expenditures on clothing for a second child to be lower than those for the first child, or the fixed costs associated with preparing a meal for, say, five versus four people.

²¹ The expenditures on FCSU include the value of spending funded through in-kind subsidies for SNAP, school lunches, WIC, and rental subsidies (Garner, 2010).

²² In contrast, for the OPM, the threshold is defined as three times the cost of a minimum food diet (as established in 1963); see App. D, 2-2, for more detail.

A different reference threshold value is calculated for renters, owners with mortgages, and owners without mortgages.

A “three-parameter” scale is used to adjust the reference threshold to families of differing size and composition mix, specified as follows (Short, 2001):

1. Scale for units with one and two adults = (number of adults) ^0.5
2. Scale for single parent units = (number of adults + 0.8 * first child + 0.5 * number of other children) ^0.7
3. Scale for all other families = (number of adults + 0.5 * number of children) ^0.7

The equations contain both “multipliers” (as in 0.8 and 0.5 in the second equation) and “exponents.” The 0.8 and 0.5 multipliers recognize that children do not consume as much as adults (who are assigned a multiplier of 1.0). The exponents 0.5 and 0.7 recognize that the additional costs of adding a member to the resource unit decreases with the number in the resource unit; in other words, the per-unit cost of basic needs decreases with household size. The threshold resource level for the reference family is multiplied by the resulting equivalence scale to determine the thresholds for each combination of family size and composition.

Table D2-2, displays the equivalence scaling ratios for selected family types relative to that of the reference family. So, for example, to achieve the poverty threshold, a one-parent, two-child family is assumed to require about 83 percent of the level of resources required by a two-parent, two-child family. In 2015, the reference renter family is assigned an SPM threshold of $25,583, estimated using the CE.²³

Table D2-3, provides a comparison of the SPM equivalence scale to the equivalence scales implicit in the three government benefit programs: the EITC, SNAP, and the Child Tax Credit (CTC).

APPENDIX D, 2-5
COST-OF-LIVING ADJUSTMENTS IN POVERTY THRESHOLDS AND BENEFITS

The cost of maintaining a given standard of living changes over time and varies from place to place. In the case of the former, prices may rise (or fall) from one period to the next, meaning that an individual or household requires more (or less) nominal income to purchase a similar “basket” of goods and services. Regarding change by place, the cost of purchasing a

___________________

²³ This is the figure for renters—the threshold for owners with a mortgage is about the same, and the threshold for owners without a mortgage is quite a bit lower than the $25,583 figure.

TABLE D2-2 SPM Equivalence Scales by Household Size, 2015

SOURCE: Fox (2018).

similar market basket may vary from one city, state, or country, to another. Cost-of-living adjustments (COLAs) are used as a method to equate dollar amounts, in terms of purchasing power, either temporally or spatially; they are often applied to payments such as those made for wage contracts.

However, with the exception of SNAP, none of the cash or near-cash benefits paid to low-income people, including the EITC and the CTC, contains any set of regional COLAs. The benefits are the same nominal amount across the entire nation. The only exception is public housing allowances, which are implicitly tied to the cost of rentals and which vary widely across the nation. In this appendix section, we explore these differences in the treatment of COLAs on the threshold side and the benefit or income side of income-based poverty measures; we also describe how the OPM and SPM address COLAs.²⁴

The Role of COLAs in Setting Poverty Thresholds

While the income thresholds (the boundary designating who is and is not living in poverty) established by the Census Bureau’s OPM are updated to account for price inflation over time (using the CPI-U), they do not include adjustments to account for geographic differences in the cost of living. As a result—assuming the measurement goal is to provide an accurate perception of the relative economic well-being of populations across the country—in high-cost states the OPM undercounts the number of people living in poverty, and in low-cost states it overcounts them, relatively speaking.

By contrast, the income thresholds set by the SPM are designed to incorporate changes in the standard of living over time and are also

___________________

²⁴ SNAP contains a modest adjustment for differences in housing costs across areas by allowing for deductions (against earned income) for shelter cost.

TABLE D2-3 Implied Equivalence Scales for EITC, CTC, and SNAP Programs by Household Size

SOURCE: Committee-generated, using data from Tax Policy Center (2017).

partially adjusted to reflect geographic differences in families’ living costs. The SPM income thresholds are based on a measure of resources required to purchase necessities—food, clothing, shelter, and utilities—at a basic level, as estimated using the previous 5 years of CE data from the Bureau of Labor Statistics.²⁵ They are re-estimated every year instead of being adjusted for inflation. The SPM thresholds generally, but not always, show a greater year-to-year increase than the OPM thresholds, indicating that living standards are outpacing inflation.

The geographic COLAs in the SPM compensate for differences in the price of rental housing, as measured by the median rent index, across areas. The median rent index is the ratio of the median outlays by renters for rent and utilities (for a two-bedroom unit with complete kitchen and plumbing facilities) in a specific metropolitan area or state to the median outlays nationwide for the same type of unit (Renwick, 2018). Rental price data from the Census Bureau’s American Community Survey are used to adjust the housing component of the poverty thresholds.

The impact on poverty threshold levels of including regional COLAs—whether based only on housing costs or on consumption items more broadly—turns out to be quite significant in terms of the resulting distribution of the population on either side of the line. It follows that incorporating geographic variation into poverty guidelines used in determining eligibility for public-benefits programs would have a considerable impact on the number of families eligible in different parts of the country (the overall number eligible nationwide might not vary much, if at all).²⁶ In high-cost areas, such as the urban areas of the East and West coasts, COLAs increase the size of the population falling below the poverty line, both in absolute terms and proportionally. Meanwhile, in lower-cost regions, such as states in the South and Midwest, the portion of the population falling below the poverty line decreases (Curran et al., 2008). For example, based on Census Bureau data for 2015–2017, Mississippi’s official poverty rate for the total population of 19.5 percent was more than 3 percentage points higher than the cost-of-living adjusted SPM for the same period. In California, during the same period, the SPM rate was 5.6 percentage points higher than the OPM rate, and at 19 percent was the second-highest rate in the nation (Fox, 2018).

Alternative Approaches to Geographic COLAs

An alternative geographic COLA approach, explored in Renwick et al. (2014), involves applying state and metropolitan regional price parities (RPPs), which account for cross-area variation in a broader set of essential consumption items rather than simply housing costs. This method may draw from either an “all item” index that tracks prices from a broad group of expenditure classes, or from an index focused on food, clothing, and rents. The Bureau of Economic Analysis has been measuring variation in living costs through its RPP program for several years (Bureau of Economic Analysis, 2017). Its estimates (which combine CPI data for various consumption expenditure classes, including rents, food, apparel, transportation, housing, education, recreation, medical, and other goods and services) can be used to express price levels for states and metropolitan areas in comparison to the overall national averages. Using the RPPs produced by the Bureau of Economic Analysis, the Tax Foundation (2017) demonstrated that, relative to what can be purchased at the national level for $100, a market basket can be purchased worth $116.01 in Mississippi, worth $115.21 in Alabama, and worth $114.42 in Arkansas. At the other end of the spectrum, $100 is effectively worth only $84.18 in Hawaii, $85.47 in the District of Columbia, and $86.73 in New York State.

Renwick et al. (2014, p. 2) found significant differences between poverty thresholds adjusted by the rent index only and those adjusted by an all-item RPP, “resulting in higher poverty rates for 15 states and lower rates for 26 states.” Even when the narrower (food, clothing, and rent) RPP COLA was used, poverty estimates were “higher than the median rent index poverty rates in 20 states, lower in 22 states and not statistically different in 9 states.” In metropolitan areas, use of the RPP lowers the poverty rates when compared to the median rent index, because percentage differences in the combined price level of goods and services are generally not as large as those for rents alone.

Both the 1995 National Research Council report on measuring poverty²⁷ and a report by an Interagency Technical Working Group (2010) concluded that although adjusting the entire market basket may be desirable for an SPM, data on price differences for elements other than rent and utilities were inadequate to do so. However, given the subsequent work on RPPs by the Bureau of Economic Analysis’s Regional Price Branch, the situation has changed such that a COLA could be implemented.

___________________

²⁷Measuring Poverty: A New Approach. Available at: https://www.nap.edu/catalog/4759/measuring-poverty-a-new-approach.

The Role of COLAs in Setting Benefits

Just as it makes sense to adjust poverty income/resource thresholds to reflect regional variation in the cost of achieving a given standard of living, it is also reasonable to consider treating benefit payment formulas similarly. Currently, most anti-poverty programs do not feature COLAs that would formulate variations in payment levels across regions. The CTC and the EITC are examples of programs that fall into this category, making them much more valuable in lower-cost areas (Fitzpatrick and Thompson, 2010).

Likewise, eligibility for SNAP, administered by the U.S. Department of Agriculture, is determined by a uniform national standard with maximum benefit allotments, deductions, and income eligibility standards (by family size) adjusted for price inflation over time (COLAs take effect on October 1 each year). With only a couple of exceptions, however, SNAP benefits to low-income families are not distributed according to the cost of living in a city or metropolitan area (USDA, 2013). Net monthly income (eligibility) limits, set at 100 percent of the poverty level for the household size, are different (higher) only for Alaska and Hawaii, and maximum benefit amounts only vary for Alaska, Hawaii, Guam, and the U.S. Virgin Islands.²⁸

Regional COLAs for food have less impact than COLAs for housing, because grocery costs vary by region considerably less than rental costs do. For example, the Council for Community and Economic Research survey estimates that average housing costs in Tulsa are only 66 percent of the national average, while in Baltimore and San Francisco they are 155 percent and 295 percent, respectively. But even in the case of food, there are some extreme differences. In Manhattan, for example, costs are 158 percent of the national average, while food costs in Tulsa are 81 percent of the national average (USDA, 2013). In either expenditure category, however, measured child poverty rates would be increased (relative to the OPM) in places like New York and California by including COLA adjustments to offset high housing or food costs.

APPENDIX D, 2-6
DIFFERENCES BETWEEN THE RESOURCE MEASURES USED BY THE OPM AND SPM POVERTY MEASURES

As noted in the text (see also Appendix D, 2-2), family resources in both the OPM and SPM poverty measures 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

___________________

²⁸ Families with housing and utility costs that exceed one-half of net income are allowed a deduction for excess shelter costs, which may be more likely to occur in areas with higher-than-average housing costs generally.

between the OPM and SPM resource definitions is that SPM-based family resources also include “near-cash” income benefits such as SNAP (formerly known as food stamps) and housing subsidies, as well as near-cash benefits from many smaller programs. The SPM resource measure is also an after-tax measure, including deductions for payroll and federal and state income taxes as well as additions to resources through the EITC and the CTC. Table D2-4 provides a more complete accounting for the differences between OPM resources and SPM resources.

APPENDIX D, 2-7
POVERTY AMONG AMERICAN INDIAN AND ALASKA NATIVE CHILDREN

American Indians and Alaska Natives (AIAN) make up a small but rapidly expanding proportion of the U.S. population (Norris, Vines, and Hoeffel, 2012).²⁹ Of the 5.3 million people who identify as AIAN, nearly one-half also identify as some other race, and the mixed-race population is growing faster than the AIAN-alone group (Norris, Vines, and Hoeffel, 2012). Because of the AIAN population’s relatively small size, data documenting it are scarce, particularly concerning AIAN children. In this appendix, we draw upon data from large-scale data collections efforts conducted by the U.S. Census Bureau.

Unlike other minority groups in the United States, the AIAN population is also recognized as a political group, with political rights that may or may not align with racial or ethnic designations. As such, the AIAN population is eligible for certain programs and benefits that would otherwise be deemed illegal or unconstitutional in other settings (e.g., preferential hiring, treaty payments, and sovereign immunity). These benefits accrue directly as a result of their unique political status and not from a racial or ethnic designation. In the discussion that follows, we will primarily discuss the AIAN population as a racial and ethnic group; in Chapter 7, we discuss potential programs that may be of benefit only to the AIAN population that are considered citizens of their tribal nations.

As measured by the OPM,³⁰ the child poverty rate among the AIAN population as a whole was 31 percent in 2015, but there are differences by race and geography (Table D2-5). The OPM poverty rate for the entire population identifying as at least part AIAN increased from 27 to 31 percent from 1990 to 2016. However, for the AIAN-alone child population,

___________________

²⁹ This appendix is adapted from a study (Akee and Simeonova, 2017) commissioned by the committee for this report.

³⁰ Poverty data measured by the Supplemental Poverty Measure for this population were not available for all time periods covered in this analysis.

TABLE D2-4 Difference Between OPM Resources and SPM Resources

NOTES: Market income poverty is a subset of either OPM or SPM poverty that researchers use when they want to compare the effects of market income on poverty separately from other income sources. AFDC/TANF = Aid to Families with Dependent Children; EITC = Earned Income Tax Credit; LIHEAP = Low-Income Home Energy Assistance Program; OPM = Official Poverty Measure; SPM = Supplemental Poverty Measure; SSDI: Social Security Disability Insurance; SSI = Supplemental Security Income; TANF = Temporary Assistance to Needy Families.

SOURCE: Adapted from Bitler, Hoynes, and Kuka (2017).

TABLE D2-5 Child Poverty Rates by Year and Population

SOURCE: Adapted from a study by Akee and Simeonova (2017) commissioned by the committee for this report.

poverty rates were higher in 1990 and then, over the 1990s, dropped by 6 percentage points (from 38 to 32%), remained constant through the 2000s before increasing to 47 percent in 2015. In comparison, Black (or African American) children experienced poverty rates that fell and then rose in roughly similar ways throughout this time period, suggesting that there were national trends that affected poverty in these different groups in approximately similar magnitudes.

The rate of poverty among AIAN children also varies by geography. One-fifth of the AIAN population lives on reservations, in traditional homelands, or in Alaska Native villages (Norris, Vines, and Hoeffel, 2012). Among the on-reservation child population, the poverty rates are on average about 10 percentage points higher at all points in time than the off-reservation AIAN population (Table D2-5). In addition, there is much less difference in the child poverty rates between the AIAN-alone and AIAN-in-combination populations, which may be due to the relatively few mixed-race AIAN children residing on reservations (Norris, Vines, and Hoeffel, 2012).

There was a larger (in percentage points) reduction in poverty for the on-reservation population than for the AIAN population residing in the United States as a whole between 1990 and 2000, which coincides with the era of widespread expansion of American Indian casino operations. Thus, while AIAN children have historically suffered from high poverty rates, there is evidence that policies expanding resources available to tribal governments have reduced poverty, at least in the tribal-enrolled population.

APPENDIX D, 2-8
THE CHANGING DEMOGRAPHY OF CHILDREN, INCLUDING CHILDREN IN POVERTY

The U.S. population is becoming increasingly racially and ethnically diverse, and the child population is even more diverse than the total population. In 2016, 51 percent of the child population was White,³¹ compared to 61 percent of the population ages 18–64, and 77 percent of the population ages 65 and over (U.S. Census Bureau, 2018a). Figure D2-1, shows the historical and projected racial/ethnic composition of the child population. Over the coming decades, the child population will become even more diverse (U.S. Census Bureau, 2018b). For example, Hispanic children represented only 9 percent of the child population in 1980 but 25 percent in 2017, and they will represent 32 percent in 2050. By midcentury, racial/ethnic “minority” children will be 61 percent of the child population (U.S. Census Bureau, 2018b), and the U.S. population as a whole is projected to become majority-minority (U.S. Census Bureau, 2018b).

The changes in the racial/ethnic composition of the child population reflect increases in immigration as well as higher fertility among minorities and especially among immigrants (Martin et al., 2018; U.S. Census Bureau, 2016). The proportion of the child population in immigrant families (those where at least one of the parents and/or the children is foreign-born) grew

___________________

³¹ Note that all race categories used in this report exclude Hispanics.

from 6 percent in 1970 to 25 percent in 2016 (Capps and Fortuny, 2006; Federal Interagency Forum on Child and Family Statistics, 2017). Children in immigrant families are more than 25 percent of the child population in 13 states and in 31 of the 100 largest metropolitan areas (Urban Institute, 2018). Child poverty in families where parents are U.S.-born is 9.9 percent, compared with 20.9 percent in families where at least one of the parents is an immigrant (i.e., foreign-born) (refer to Table D2-5). Child poverty is even higher in households where some members are not citizens or are authorized (refer to Table D2-5).

The increasing diversity of the child population—driven largely by the growth in the Hispanic child population—coupled with higher poverty rates among Hispanic, Black, and AIAN children has led to significant changes in the composition of the child population in poverty. As shown in Figures D2-2 and D2-3, the increasing diversity of children in poverty is apparent in both OPM-based and SPM-based figures. In 2016, 33 percent of the poor child population (<100 percent SPM) was White, down from 55 percent in 1970, while in 2016 Hispanic children represented 40 percent of the population of children in poverty, up from 12 percent in 1970. Hispanic children have been the largest group of children in poverty (4.6

million in 2016) since 2002, followed by White children (3.4 million) and Black children (2.5 million). The trends in the composition of children in poverty based on the OPM are similar to those discussed above based on the SPM, although based on the OPM measure Hispanic children overtook White children to become the largest group in poverty in 2007, 5 years later than they did following the SPM measure.

The downward trend in child poverty for all groups, as measured in SPM-based rates, is readily apparent in Figure D2-4. Because the decline was steeper for minorities, racial/ethnic gaps in child poverty also declined during this period. Between 1970 and 2016, the absolute difference in poverty rates between Black children and White children declined from 38 to 16 percentage points, and between Hispanic and White children from 32 to 16 percentage points. In contrast, gaps in OPM-based poverty rates declined by only about 6 percentage points between Black and White children and remained constant, at about 20 percentage points, between Hispanic and white children (Figure D2-5). The difference between trends

in racial/ethnic gaps computed using SPM and OPM reflects the growth of in-kind transfers to the poorest families, who were disproportionately minority.

Additionally, family structure has changed dramatically in recent decades and, as shown in Figure D2-6, poverty rates have declined for each group—single parent, cohabitating parents, and married parents. TRIM3-adjusted poverty rates for select demographic groups (defined by age of child, region, metropolitan/nonmetropolitan status, disability, and health insurance status) for 2015 are shown in Table D2-6, for deep poverty, poverty, and near poverty.

APPENDIX D, 2-9
DISTRIBUTION OF CHILD POPULATION ACROSS PERSISTENTLY HIGH-POVERTY COUNTIES

Child poverty rates vary greatly not only by child demographic characteristics but also geographically. The material hardships associated with poverty and families’ ability to get out of poverty vary across communities. The experience of child poverty in a community with good schools and resources for families and with pathways for economic mobility may be different than in a community that has suffered persistent poverty for decades. The burden of poverty may be harder in rural areas where access to services for low-income families may be limited (Schaefer, Mattingly, and Johnson, 2016).

As discussed in Chapter 2, we classified a county as experiencing persistently high poverty if 20 percent or more of children under 18 years old were poor as measured by the 1980, 1990, and 2000 decennial censuses and the American Community Survey’s 5-year estimates for 2007–2011.³² Using this definition, we classified 708 of 3,141 counties in the United States as persistently high-poverty counties. The analyses use county-level estimates of the U.S. child population by race/ethnicity provided by the Census Bureau as part of the U.S. Population Estimates Program (data are as of July 1, 2015). In analyses by race/ethnicity, Hispanics may be of any race; other racial groups exclude Hispanics.

In the following, we analyze the distribution of the child population across persistently poor and nonpoor counties, focusing on disparities between racial/ethnic groups, metropolitan and nonmetropolitan areas, and states. We then repeated the analysis using point-in-time poverty rates

___________________

³² This definition was adapted from the U.S. Department of Agriculture. For more information, see https://www.ers.usda.gov/webdocs/DataFiles/48652/ERSCountyTypology2015Edition.xls.

to classify counties as poor versus nonpoor. Specifically, we used the 2015 poverty rate for children under 18 from the Census Small Area Income and Poverty Estimates (SAIPE) Program to classify counties as poor if 20 percent or more children under 18 years old were poor in 2015. Using this definition, we classified 1,858 of 3,141 counties as poor counties. By this definition, more than one-half of all counties are currently poor. We refer to either persistently poor counties or point-in-time poor counties to distinguish between the two definitions.³³

Approximately 10 million children ages 0 to 17 resided in persistently poor counties in 2015 (Table D2-7). This corresponds to 13.9 percent of the 74 million children under age 18 living in the United States. Of children living in persistently poor counties, 73 percent resided in metro areas, while 27 percent resided in nonmetro areas (Table D2-6). There are significant racial/ethnic differences in the proportion of children who live in persistently poor counties: from 8.2 percent of Asian and Pacific Islander children to 36 percent of AIAN children (Figure D2-7). In all racial/ethnic groups, except for AIAN children, the majority of children in persistently poor counties live in metropolitan areas (Figure D2-8). However, children in persistently poor counties are more likely to live outside metropolitan areas (26.7%) than children in nonpersistently poor counties (11.9%) (Table D2-8). As discussed in greater detail later in this appendix, most children in persistently poor counties live in the South (6.2 million, 61.1 percent, see Table D2-11). In absolute terms, (non-Hispanic) White children represent the largest racial/ethnic group of children living in persistently poor counties (Table D2-8). However, compared to the 3.6 million White children living in persistently poor counties, the number of Hispanic and (non-Hispanic) Black children living in such counties is of similar magnitude, 3.1 million and 2.7 million, respectively.

Among White children residing in persistently poor counties, 61 percent live in metro areas (Figure D2-8). The comparable figures for Black, Hispanic, and AIAN children are 77.7 percent, 85.4 percent, and 22.6 percent, respectively. In nonmetro area counties that are persistently poor, 51 percent of children are White, 22.4 percent are Black, and 16.6 percent are Hispanic (Figure D2-9). In metro-area counties that are persistently poor, each of these groups contributes about 30 percent of the child population (Figure D2-9).

In contrast, 66 percent of children living in currently poor metro counties are White, 12.1 are Black, and 13.8 are Hispanic (Figure D2-10). In

___________________

³³ We also repeated parts of the analysis using the 2015 poverty rate of the total population to classify counties as poor vs. nonpoor. In this case, counties were classified as poor if 20 percent or more of the total county population was poor. Using this classification, 755 of 3,414 counties were classified as poor.

TABLE D2-6 Percent of Children in Poverty based on TRIM3—Adjusted SPM for 2015 by Level of Poverty

^a The Other (Non-Hispanic) race/ethnicity category includes children who are Asian or Pacific Islanders or American Indian or Alaska Natives, or who report more than one race.

^b The Metro status category categorizes a child’s household by geography. A metropolitan area includes a large population nucleus and any outlying communities that are highly socially or economically integrated with the nucleus. Nonmetropolitan areas are those outside of metropolitan areas. Households in the Principal city of metro area category live within the largest city of the metropolitan area, and Metro area, nonprincipal city captures households in communities adjacent to the largest city. Some households do not have their metro status suppressed to preserve privacy, and are captured in the Metro Area, principal city status not disclosed category. For more information on this category, see the Current Population Survey Subject Definitions, available at https://www.census.gov/programs-surveys/cps/technicaldocumentation/subject-definitions.html.

^c If a child has at least one biological, adoptive, or step-parent that was born in another country, the child is classified as having an immigrant parent. Persons born abroad to American parents are counted as native-born. If a child does not have a parent present, the immigrant status of the unit head and the unit head’s spouse are used, unless the child is the unit head, spouse, or cohabiting partner (these children are tabulated separately).

^d The child is placed in the first row that applies. Citizens include both native-born and naturalized citizens. An Other noncitizen includes noncitizens not classified as unauthorized, recent, or temporary. These include legal permanent residents who have been in the United States for more than 5 years and refugees/asylees.

^e These rows reflect the work status of persons aged 18 or older in the unit. Full-year is classified as 50 weeks or more, and full-time is classified as 35 hours per week or more.

^f A child is classified as being in a unit with a person with a disability if there is at least one person in the unit who is younger than 65 and is identified as disabled according to the definition used when determining SSI eligibility.

^g These rows reflect the characteristics of the biological, adoptive, or stepmother (if present), otherwise the biological, adoptive, or stepfather (if present). If neither the father nor mother is present, then the characteristics of the SPM unit head are used, unless the child is the unit head, spouse, or cohabiting partner.

SOURCE: TRIM3 analyses commissioned by the committee.

currently poor non-metro counties, those groups make up 39.1, 19.3, and 33.1 percent of the child population, respectively (Figure D2-10). At 39.8 million, the number of children residing in point-in-time poor counties in 2015 was considerably larger than the number in persistently poor counties, which is not surprising given the much larger number of counties that are poor at a point in time (Table D2-8). Persistent poverty is more prevalent outside of metropolitan areas than point-in-time poverty. While 27 percent of children residing in persistently poor counties live outside a metro area, only 18 percent of children residing in currently poor counties do (Tables D2-7 and D2-8).

The racial/ethnic composition of children living in persistently poor counties differs from that of children living in point-in-time poor counties (see Tables D2-9 and D2-10). While white children accounted for 35 percent of children living in persistently poor counties, they make up 44 percent of poor children in point-in-time poor counties (Tables D2-8 and D2-9). Relative to the total number of children of a given race/ethnicity, the risk of residing in a point-in-time poor county is highest among Black children (70.8%), followed by AIAN (70.6%), Hispanic (65.0%), and (non-Hispanic) White children (46.0%) (Figure D2-7).

The percentage of children under 18 living in persistently poor counties varies across U.S. regions. In the South, 22.1 percent of children live in persistently poor counties. In the Northeast, the corresponding figure is 17.3 percent (Figure D2-11). Together, the South and Northeast account

for most of the child population living in persistently poor counties (Table D2-11). Of the 10.2 million children living in persistently poor countries, 81.2 percent reside in the Northeast and South (Table D2-11). In 10 states and the District of Columbia, more than a quarter of the child population lives in persistently poor counties (Figure D2-13). In Mississippi, nearly two-thirds of the child population live in persistently poor counties. In Alabama, Louisiana, and New Mexico, more than 40 percent of the child population live in persistently poor counties (Figure D2-12).

While there are a large number of White children living in persistently poor counties across the United States (Figures D2-12 and D2-13), AIAN, Black and especially Hispanic children residing in persistently poor counties are more geographically concentrated. Black children in persistently poor counties reside in certain states in the South and Northeast, especially Alabama, Georgia, Louisiana, Mississippi, and New York. Hispanic children in persistently poor counties reside especially in California, New York, and

TABLE D2-8 Distribution of Children Under 18 Across Point-in-Time Poor and Nonpoor Counties, by County Metro Area Status

SOURCES: U.S. Population Estimates, 2016 Vintage, Census Bureau. Data as of July 1, 2015. 2015 county poverty rates from Census Small Area Income and Poverty Estimates (SAIPE) Program data.

TABLE S2-9 Distribution of Children Under 18 Across Persistently Poor and Nonpoor Counties, by Race and Ethnicity

SOURCE: U.S. Population Estimates, 2016 Vintage, Census Bureau. Data as of July 1, 2015. Hispanics may be of any race. Other racial groups exclude Hispanics.

Texas, which jointly account for 70 percent of Hispanic children living in persistently poor counties; Texas alone accounts for 43.5 percent. Arizona, New Mexico, North Carolina, and Oklahoma account for 60 percent of Native American children living in persistently poor counties.

TABLE D2-11 Child Population Living in Persistently Poor and Currently Poor Counties, by Region

SOURCES: U.S. Population Estimates, 2016 Vintage, Census Bureau. Data as of July 1, 2015. 2015 county poverty rates from Census Small Area Income and Poverty Estimates (SAIPE) Program data.

APPENDIX D, 2-10
ANCHORED AND UNANCHORED METHODS OF CALCULATING SPM POVERTY OVER TIME

Changes in poverty measured using the SPM could be in part due to changes in poverty thresholds. The SPM was designed to be a relative measure of poverty with poverty thresholds increasing over time as living standards at the 33rd percentile of the American income distribution increased. Some observers such as Wimer et al. (2013, 2016) and Fox et al. (2015) have suggested that an anchored measure may be more useful for measuring poverty over time because it applies a single threshold over the entire time period. An anchored measure of poverty does not take into account changes in living standards when assessing changes in poverty over time (Wimer et al., 2013). Rather, it uses a fixed benchmark for living standards, making it arguably more useful for establishing how the amount of families’ resources has changed over time. Like the OPM, the anchored SPM is an absolute measure of poverty. (Also see discussion of absolute versus relative poverty measures in Appendix D, 2-2.)

The anchored SPM used in the text backdates and updates the poverty threshold from 2012 to reflect today’s consumption norms while adjusting only for inflation. The result is an absolute poverty measure that shows poverty trends over time in relation to contemporary consumption patterns and inflation. Figure D2-15 shows the difference in trends between the anchored and unanchored or absolute and relative SPM measures.

The relative, unanchored SPM shows less progress against child poverty than the absolute, anchored SPM poverty measure. This is because

living standards at the 33rd percentile of the income distribution increased somewhat more rapidly than inflation. The poorest families with children were better off in an absolute sense—the anchored poverty rate dropped by nearly one-half. But when compared to the progress of lower middle class families—those at the 33rd percentile—the progress of the poorest families is less impressive—only a 25 percent cut in poverty.

As described in the text, the committee chose to focus on the anchored poverty measure because it wished to isolate the effects of income transfers from the effects of changes in living standards. This is especially salient for comparisons of the OPM to the SPM. A comparison of the OPM to unanchored SPM mixes together differences in counting key transfers with differences between an absolute and relative poverty measure. Whether the absolute or relative SPM is a superior measure remains an open question. For further discussion see Wimer et al. (2013, 2016) and Garfinkel, Rainwater, and Smeeding (2010). Future research should address this issue by comparing the two time series with other measures of economic deprivation such as food insecurity and other forms of material hardship.

APPENDIX D, 2-11
POVERTY MEASUREMENT ACROSS COUNTRIES: CROSS-COUNTRY POVERTY LINES AND CHILD POVERTY RATES

To understand child poverty in the United States relative to peer nations, the committee commissioned analyses comparing child poverty measures in the United States to four Anglophone nations: Australia, Canada, Ireland, and the United Kingdom. Data from the Luxembourg Income Study (LIS) and the Organisation for Economic Co-operation and Development (OECD) poverty and income database were used to compare child poverty across the five nations. The results of these analyses are discussed in Chapter 2.

The common international standards and measurement methods for poverty statistics are different from those of the SPM measure in several ways. For this reason, comparisons of poverty across countries are problematic. Both LIS and OECD calculate poverty based on disposable income. The SPM definition of disposable income largely follows the LIS and OECD but differs in a number of ways. Specifically, it

• subtracts medical out-of-pocket expenses and work-related costs;
• adjusts for cost-of-living differences within the nation;
• separates poverty lines for owners with and without mortgage and renters; and
• uses cohabiter units and families who meet income-sharing rules, not households.

The SPM also takes a different approach to equivalence scales. Poverty measurement typically adjusts income for family size using an equivalence scale (as described in Appendix D, 2-1), reflecting a less-than-proportional increase in expenses as the number of family members increases. For example, considering the resource needs of a family with a given number of children, one can express the spectrum of possible adjustments through the exponent X in the following simple equation:

Needs = (Number of children)^X

where X = 1 in the case of equal needs for all family members and X = 0 if needs do not increase at all with increases in family size. As detailed in Appendix D, 2-1, the SPM sets X at 0.7 for families with children, which effectively assumes that having two children generates the need for an income that is 62 percent greater than having one child, and that having three children generates 2.2 times the need of one. The equivalence scale in the LIS sets X at 0.5, which differs from both the OECD scale and the

SPM scale. Finally, the LIS and OECD statistics do not include adjustments for underreporting for benefits or other income sources.

Cross-Country Relative Poverty

Figure 2-13 (in the body of Chapter 2) first presents child poverty rates using the OECD-50 for the United States and the four anglophone comparison countries in 2015. This poverty measure indicates that the child is poor if the equivalized income is less than one-half of the median. The United States is the clear outlier in terms of relative poverty rates: 19.9 percent of children in the United States are classified as poor compared to 15 percent in Canada, 13 percent in Australia, and less than 12 percent in the United Kingdom and Ireland according to OECD-50. The qualitative results compare well with similar measures from LIS (see LIS-50 in Figure D2-16) as well as the existing literature (e.g., Smeeding and Thevenot, 2016; Gornick and Nell, 2017).

To understand differences in child poverty across countries with an SPM-type poverty threshold, the committee attempted to adjust the SPM to the international standards. To construct an SPM-based relative poverty measure, we took the four-person U.S. SPM poverty threshold and compared it to the median income for a family of four in the United States in the LIS. Ignoring health care costs and work expenses and other adjustments, the SPM poverty line cuts the LIS U.S. income distribution at the 40th percentile. Based on this, we then define relative poverty lines across all five countries at the 40th percentile of their respective income distributions. The resulting child poverty rates, shown as LIS-SPM-40 in Figure 2-13 in Chapter 2 and Figure D2-16, produce a country ordering similar to that based on the OECD poverty measures, with U.S. poverty rates (12.3%) much higher than those in all other countries, including the second- highest ranked, Canada (10.1%).

Cross-Country Absolute Poverty

In addition to relative poverty comparisons, it is also useful to compare countries based on absolute poverty differences. It is not possible to implement all of the SPM adjustments in LIS data, nor can the OECD data be changed to accommodate an absolute poverty line. Instead, the committee adjusted the SPM to be comparable to the LIS. Building on the analysis above, the SPM poverty line, ignoring health care costs, work expenses and other adjustments for COLAs and housing status, is about 40 to 41 percent of U.S. median adjusted income (Short, 2013; Wimer and Smeeding, 2017). The committee used these and set the poverty line for two adults and two children as $25,551, which was 40.2 percent of LIS median household

income (bottom panel of Figure D2-16). The LIS team converted this line to other nations’ incomes using purchasing power parities (PPPs) and national price indices where years differ (Gornick and Jantti, 2013; Gornick and Nell, 2017; Rainwater and Smeeding, 2003).³⁴

Using these PPP-adjusted poverty thresholds, we can compare child poverty rates in the five countries in terms of absolute poverty; this is shown in Figure D2-16 and labeled LIS-SPM-PPP. Reflecting higher average income in the United States compared to our peer English-speaking nations, using an absolute poverty line across countries rather than a relative poverty line within each country leads to a different cross-country ranking. In particular, as shown in the bottom panel of Figure D2-16, the absolute poverty line is about 40 percent of median income in the United States, Canada, and Australia, but it is 52 percent of median income in Ireland and almost 56 percent of median income in the United Kingdom. The resulting absolute child poverty rates (fourth panel of Figure D2-16, LIS-SPM-PPP) are highest in the United Kingdom (13.5%), followed by the United States (12.5%), Ireland (11.3%), and Canada (10.3%)—and are quite a bit lower in Australia (8.1%).

We see where the LIS-SPM-PPP poverty line cuts the distributions in other nations at the bottom of Figure D2-16. In contrast, we could take the official U.S. poverty line used by others to make absolute comparison across nations (Gornick and Nell, 2017) and see where it cuts the other nations’ distributions. The comparison of these approaches for the United States and United Kingdom indicates the sensitivity of absolute poverty measures in an international context. In Figure D2-16 we compare the two absolute poverty lines for a family of four (both in PPP 2013 prices): the $25,551 LIS-SPM-PPP poverty line and the $23,306 U.S. official poverty line.

We converted the lines to adjusted income per adult equivalent to assure the same LIS square-root equivalence scale in both comparisons (Figure D2-17). This resulted in equalized poverty lines of $12,348 per equivalent adult for the SPM PPP line and $11,518 per equivalent adult for the U.S. official line. These lines are drawn vertically and compared to the cumulative distributions of real PPP adjusted income per equivalent adult for families with children in the United States (blue) and the United Kingdom (orange). At the U.S. official line, child poverty is slightly higher in the United States, at 10.5 percent, than in the United Kingdom, at 10.0 percent. But at the SPM line, the opposite is true: the U.S. child poverty

___________________

³⁴ The 2013 U.S. SPM translates into about $25,550 for two parents and two children. These are converted to other currencies first using 2011 PPP and then using national consumer price changes when the LIS years differ from 2013. See Pinkoyskiy and Sala-i-Martin (2018) for more information about using PPPs.

rate of 12.5 percent is below the UK child poverty rate of 13.5 percent, as shown in Figure 2-13 in Chapter 2 and Figure D2-16.³⁵

Using these LIS-based absolute poverty measures, we can extend this analysis to make a fuller comparison of poverty rates for the United States and the four comparison countries. Figure D2-18 compares absolute poverty at the deep poverty line (one-half the U.S. SPM-PPP line in absolute terms) and near poverty (using 150% of the U.S. SPM-PPP poverty line). As the figure shows, the highest rates of absolute deep poverty are in the United States (3.6%, far above the next highest, which is in Australia at 1.9%) but the highest rates of near-poverty are found in the United Kingdom (46.4%).

Figure D2-19 shows the absolute child poverty rates among single-parent families, families with one or more full-time worker, and immigrant

___________________

³⁵ These results are comparable to those found in Gornick and Nell (2017), who used 2011 data to make the same calculations.

families using the same absolute measure (LIS-SPM-PPP). Children in single-parent families are considerably poorer in the United States (30.6%) than in other nations in absolute poverty terms, even comparing the United Kingdom with its lower living standards as well as Australia (19.1%) and Canada (26.3%) with similar living standards. In contrast, child poverty rates in two-parent families are only 6.4 percent in the United States. Only Australia has a lower rate, at 5.9 percent (not shown here). But the low-wage United States has the second highest absolute poverty rate for children in families with at least one full-time worker, 7.0 percent, with only

the United Kingdom higher, at 8.8 percent, while all the other nations fall below 4 percent (Figure D2-19).

Another important dimension to compare across countries is with respect to immigrant status (bottom panel of Figure D2-19). However, in the LIS, comparable data on immigrant status is available only for Australia

and Ireland.³⁶ In these two countries, absolute immigrant child poverty rates are below 8 percent, compared to 14.4 percent in the United States. Finally, the LIS estimated absolute child poverty excluding non-Hispanic Blacks in the United States, lowering the U.S. absolute poverty rate from 12.5 to 10.3 percent. Racial differences therefore make an almost 20 percent difference in the U.S. poverty rate compared to other English-speaking rich nations without a significant racial minority.

Cross-Country Comparisons of Income

Given the differences in country rankings of absolute and relative poverty measures, we conclude this appendix with a discussion of disposable household income and other real income concepts to get a better understanding of living standards across the five English-speaking nations of interest. In the LIS results, different equivalence scales, different years and different PPPs may all produce different results in terms of median incomes. Further, there is underreporting in all household income surveys. These are corrected for in the TRIM3 simulations, but not in LIS or OECD surveys.

And so, here we explore differences in living standards across countries by comparing three series of real PPP adjusted income indexes to the United States. These results are shown in Table D2-12. The first column shows the LIS-SPM-PPP poverty lines as a fraction of median incomes, as seen in Figure D2-16. The second column shows ratios of median equivalized income from the LIS database. The third column shows OECD household disposable income per capita (adjusted to National Income and Product Account totals). And the fourth column shows GDP per capita from the International Monetary Fund. For each measure, the four comparison countries are presented relative to the United States (so the value for the United States equals 100 for each).

There are several things to take away from this table. First, all of the numbers are less than 100 for each of the four comparison countries, indicating that on a range of measures the United States is on average richer than our peer English-speaking countries. Second, across all measures, Australia and Ireland come closest to the U.S. values. Third, the magnitude of the disparities across countries varies significantly across the measures.

Based on the LIS median equivalized income, Australia and Canada are nearly identical to the United States (98.7 for Australia and 99.7 for Canada). However, based on disposable income per capita or GDP per capita, the disparities are much larger. The opposite pattern is present in Ireland, although the Ireland LIS data (from 2010) are probably showing lower incomes due to the Great Recession. Finally, the United Kingdom

___________________

³⁶ The Canadian and UK LIS datasets suppress immigrant status.

TABLE D2-12 Income Comparisons Across Anglophone OECD Countries Using Different Measures of Real Income, Indexed to USA = 100

NOTES: LIS from Figure D2-16; Poverty line as a fraction of equivalized LIS median income, using 2011 PPP and domestic CPI changes to correct year; LIS: by division, Real 2013 LIS ratio of median equalized incomes in each country to USA 2013, using 2011 PPP and domestic CPI changes to correct year.

SOURCES: Analyses commissioned by the committee from the LIS Cross-National Data Center; OECD real NIPA adjusted household income per capita from https://data.oecd.org/hha/household-disposable-income.htm, using 2011 PPP; IMF Real GDP per capita from http://www.imf.org/external/pubs/ft/weo/2017/01/weodata/index.aspx, using 2011 PPP.

has the lowest real income compared to the United States using all three aggregate income measures, as shown in columns two, three, and four of Table D2-12. Hence the SPM poverty line, at 55 percent of LIS median adjusted income in the United Kingdom in the first column, reflects the lower real UK living standards (evident using three different measures).

REFERENCES

Bitler, M., Hoynes, H., and Kuka, E. (2017). Child poverty, the Great Recession, and the social safety net in the United States. Journal of Policy Analysis and Management 36(3), 358–389.

Bureau of Economic Analysis. (2017). Real Personal Income for States and Metropolitan Areas. Washington, DC. Available: https://www.bea.gov/newsreleases/regional/rpp/rpp_newsrelease.htm.

Capps, R., and Fortuny, K. (2006). Immigration and Child and Family Policy: Paper 3. Washington, DC: The Urban Institute and Child Trends.

Curran, L.B., Wolman, H., Edward, W., and Furdell, K. (2008). Poverty, Programs, and Prices: How Adjusting for Costs of Living Would Affect Federal Benefit Eligibility. Washington, DC: The Brookings Institution. Available: https://www.brookings.edu/wp-content/uploads/2016/06/0317_costofliving_curran.pdf.

Federal Interagency Forum on Child and Family Statistics. (2017). America’s Children: Key National Indicators of Well-Being, 2017. Washington, DC: Government Printing Office.

Fisher, G.M. (1992). The development and history of the poverty thresholds. Social Security Bulletin, 55(4). Washington, DC: Social Security Administration.

Fitzpatrick, K., and Thompson, J.P. (2010). The interaction of metropolitan cost-of-living and the federal Earned Income Tax Credit: One size fits all? National Tax Journal, 63(3), 419–445.

Fox, L. (2018). The Supplemental Poverty Measure: 2017. Washington, DC: U.S. Census Bureau. 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.

Garfinkel, I., Rainwater, L., Smeeding, T. (2010). Wealth and Welfare States: Is America a Laggard or Leader? New York, NY: Oxford University Press.

Garner, T. (2010). Supplemental Poverty Measure Thresholds: Laying the Foundation. Washington, DC: Bureau of Labor Statistics.

Gornick, J.C., and Jantti, M. (Eds). (2013) Income Inequality: Economic Disparities and the Middle Class in Affluent Countries. Stanford, CA: Stanford University Press.

Gornick, J.C., and Nell, E. (2017). Children, Poverty, and Public Policy: A Cross-National Perspective. LIS Working Paper No. 701. Luxembourg: LIS Cross-National Data Center. Available: http://www.lisdatacenter.org/wps/liswps/701.pdf.

Hausman, J. (2003). Sources of bias and solutions to bias in the Consumer Price Index. Journal of Economic Perspectives 17:23–44.

Hurst, E. (2012). Comment: Winning the war: Poverty from the Great Society to the Great Recession. Washington, DC: The Brookings Institution. Available: https://www.brookings.edu/bpea-articles/winning-the-war-poverty-from-the-great-society-to-the-great-recession.

Interagency Technical Working Group. (2010). Observations from the Interagency Technical Working Group on Developing a Supplemental Poverty Measure. Available: https://www.federalregister.gov/documents/2010/05/26/2010-12628/developing-a-supplementalpoverty-measure.

Johnson, N. (2017). A Comparison of PCE and CPI: Methodological Differences in U.S. Inflation Calculation and their Implications. Paper presented at the 2016 Joint Statistical Meetings of the Bureau of Labor Statistics, Chicago, IL. Available: https://www.bls.gov/osmr/pdf/st170010.pdf.

Martin, J.A., Hamilton, B.E., Osterman, M.J., Driscoll, A.K., and Drake, P. (2018). Births: Final Data for 2016. Hyattsville, MD: National Center for Health Statistics.

Meyer, B.D., and Mittag, N. (2015). Using Linked Survey and Administrative Data to Better Measure Income: Implications for Poverty, Program Effectiveness, and Holes in the Safety Net. Working Paper No. 21676. Cambridge, MA: National Bureau of Economic Research.

Meyer, B.D., and Sullivan, J.X. (2003). Measuring the well-being of the poor using income and consumption. Journal of Human Resources, 38, 1180–1220.

________. (2011). The Material Well-Being of the Poor and Middle Class Since 1980. Working Paper No. 2011-04. Washington, DC: American Enterprise Institute. Available: http://www.aei.org/wp-content/uploads/2011/10/Material-Well-Being-Poor-Middle-Class.pdf.

________. (2012a). Identifying the disadvantaged: Official poverty, consumption poverty, and the new supplemental poverty measure. Journal of Economic Perspectives 26(3), 111–136.

________. (2012b). Winning the War: Poverty from the Great Society to the Great Recession. Washington, DC: The Brookings Institution.

________. (2017). Annual Report on U.S. Consumption Poverty: 2016. Washington, DC: American Enterprise Institute.

________. (2018). Annual Report on U.S. Consumption Poverty: 2017. Washington, DC: American Enterprise Institute.

Meyer, B.D., and Wu, D. (2018). The Poverty Reduction of Social Security and Means Tested Transfers. Working Paper No. 24567. Cambridge, MA: National Bureau of Economic Research.

Meyer, B.D., Mok, W.K.C., and Sullivan, J.X. (2009). The Under-Reporting of Transfers in Household Surveys: Its Nature and Consequences. Working Paper No. 15181. Cambridge, MA: National Bureau of Economic Research.

________. (2015). Household surveys in crisis. Journal of Economic Perspectives, 29(4), 199–226.

Moffitt, R.A., and Scholz, J.K. (2009). Trends in the Level and Distribution of Income. Working Paper No. 15488. Cambridge, MA: National Bureau of Economic Research. Available: http://www.nber.org/papers/w15488.

National Research Council. (1995). Measuring Poverty: A New Approach. Washington, DC: National Academy Press.

________. (2013). Measuring What We Spend: Toward a New Consumer Expenditure Survey. Washington, DC: The National Academies Press.

Norris, T., Vines, P.L., and Hoeffel, E.M. (2012). The American Indian and Alaska Native Population: 2010. 2010 Census Briefs. Washington, DC: U.S. Census Bureau.

Pinkovskiy, M., and Sala-i-Martin, X. (2018). Shining a Light on Purchasing Power Parities. Working Paper No. 24419. Cambridge, MA: National Bureau of Economic Research.

Rainwater, L., and Smeeding, T.M. (2003). Poor Kids in a Rich Country: America’s Children in Comparative Perspective. New York, NY: Russell Sage Foundation.

Renwick, T., Aten, B., Figueroa, E., and Martin, T. (2014). Supplemental Poverty Measure: A Comparison of Geographic Adjustments With Regional Price Parities vs. Median Rents from the American Community Survey. Washington, DC: Bureau of Economic Analysis. Available: https://www.bea.gov/papers/pdf/BEAWP_PovertyMeasure.pdf.

Schaefer, A., Mattingly, M.J., and Johnson, K.M. (2016). Child Poverty Higher and More Persistent in Rural America. Durham, NH: University of New Hampshire, Carsey School of Public Policy. Available: https://scholars.unh.edu/cgi/viewcontent.cgi?referer=&httpsredir=1&article=1265&context=carsey.

Short, K. (2001). Experimental Poverty Measures: 1999. Washington, DC: U.S. Government Printing Office.

Short, K.S. (2013). The Supplemental Poverty Measure: Examining the Incidence and Depth of Poverty in the U.S. Taking Account of Taxes and Transfers in 2012. U.S. Census Bureau: Washington, DC. Available: https://www.census.gov/content/dam/Census/library/working-papers/2013/demo/kshort-sea2013.pdf.

Tax Foundation. (2017). What is the Real Value of $100 in Your State? Washington, DC. Available: https://taxfoundation.org/real-value-100-state-2017.

Tax Policy Center. (2017). EITC Parameters, 1975 to 2017. Washington, DC. Available: https://www.taxpolicycenter.org/statistics/eitc-parameters.

U.S. Department of Agriculture. (2013). Supplemental Nutrition Assistance Program, July 2013. Available: http://www.fns.usda.gov/snap/applicant_recipients/eligibility.htm#employment.

U.S. Census Bureau. (2016). Fertility: 2016 American Community Survey 1-year Estimates. Washington, DC.

________. (2018a). 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, Population Division.

________. (2018b). Race by Hispanic Origin: Main Projections Series for the United States, 2017-2060 Washington, DC: U.S. Census Bureau, Population Division.

Urban Institute. (2018). Data from the Integrated Public Use Microdata Series Datasets Drawn from the 2014 and 2015 American Community Survey. Washington, DC.

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., 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.

________. (2016). Progress on poverty? New estimates of historical trends using an anchored Supplemental Poverty Measure. Demography, 53(4), 1207–1218.

Winship, S. (2016). Poverty After Welfare Reform. New York: Manhattan Institute. Available: https://www.manhattan-institute.org/sites/default/files/R-SW-0816.pdf.

APPENDIX D, 3-1
ASSOCIATIONS BETWEEN POVERTY AND CHILD OUTCOMES

A large literature documents strong and consistent associations between child poverty and a broad range of negative outcomes for child health and well-being. In Chapter 3, the committee elected to focus on studies that estimated a causal relationship between poverty and child outcomes. In this appendix, we try to give a broader overview of some of the correlational literature, in order to show the pervasiveness of the relationship across many different types of outcomes. This appendix section is not intended as an exhaustive review of the literature. The bulk of this review is organized by domains of child outcomes into the following 10 topical sections: (1) family functioning, child maltreatment, domestic violence, and adverse childhood experiences; (2) material hardship; (3) physical health; (4) fetal health and health at birth; (5) brain development; (6) mental health; (7) educational attainment; (8) risky behaviors, crime, and delinquency; (9) the timing of poverty; and (10) severity of poverty. We also touch upon how these factors can contribute to the intergenerational transmission of poverty.

In addition to the narrative overview, we also offer summary tables of the material treated in the text, one for each section. These tables enable readers to quickly scan the findings, or to locate particular studies. This appendix ends with a consideration of the literature about the relationships between child outcomes and the timing and severity of poverty. The literature is also summarized in companion tables.

FAMILY FUNCTIONING, CHILD MALTREATMENT, DOMESTIC VIOLENCE, AND ADVERSE CHILDHOOD EXPERIENCES

A number of studies have linked child poverty to Adverse Childhood Experiences (ACEs) (Anda et al., 2006) (see Table D3-1). ACEs include abuse or neglect, the death of a parent, divorce or separation of parents, domestic violence, neighborhood violence, family mental illness or substance abuse, and incarceration of a family member. Poor and near-poor children are more than twice as likely to have experienced three or more ACEs than their more affluent peers (Anda et al., 2006; Figure D3-1). Experiencing ACEs early in life has been shown to be predictive of long-lasting negative outcomes in adulthood, such as increased risk for cardiovascular disease, obesity, smoking, drug and alcohol abuse, risky sexual behavior, and early mortality (Anda et al., 2006). It is thought that these life events cause high levels of biological stress on the developing brain, as well as on neurological, hormonal, and immune-response systems, leading to lifelong changes (Shonkoff et al., 2012). Moreover, ACEs have a “dose-response”

TABLE D3-1 Adverse Childhood Experiences

SOURCES: Health Resources and Services Administration, Maternal and Child Health Bureau; and Centers for Disease Control and Prevention, National Center for Health Statistics, National Survey of Children’s Health. Analyzed by the Health Resources and Services Administration’s Maternal and Child Health Bureau. See https://mchb.hrsa.gov/chusa14/special-features/adverse-childhood-experiences.html#sourcef2.

effect, with children who experienced larger numbers of ACEs experiencing worse outcomes (Anda et al., 2006).

More recently, researchers have suggested adding two additional ACEs: chronic economic hardship and being treated unfairly due to race/ethnicity. These nine ACEs have been added to the National Survey of Children’s Health (NSCH), allowing analysis of association of ACEs with poverty. Since poverty is now one of these ACE questions asked of parents, analysis by poverty levels excludes the experience of economic hardship. Analysis of data from the 2011/12 NSCH documents the significant association of ACEs with child poverty. There is a clear income gradient for ACEs, as shown in Figure D3-1 (National Survey of Children’s Health, 2017). We explore two ACEs, child maltreatment and intimate partner violence, below.

Researchers have repeatedly documented the association between poverty and the risk of child maltreatment, especially in cross-sectional studies of the general population. Children in low socioeconomic status (SES) families were five times more likely than those in higher income families to experience maltreatment of any kind, three times more likely to be physically or sexually abused, and seven times more likely to be neglected (Sedlak et al., 2010). In a study looking at the impact of welfare reform, higher poverty rates were associated with significantly more substantiated cases of child

maltreatment, cases of physical abuse, and cases of neglect (Paxson and Waldfogel, 2003). Almost one-half of children placed in out-of-home foster care come from homes that are eligible for welfare (Macomber, 2006).

There is some evidence that increasing family income reduces risk of child maltreatment. Cancian and her colleagues (2013) took advantage of the random assignment evaluation of child support during welfare reform in Wisconsin to distinguish the causal effect of income from other causes of child maltreatment. The experimental group received a pass-through of child support and the control group did not. Mothers received additional income, on average, of $81 to $164 per year for 2 years, although some mothers received substantially more. Despite this modest increase in income, children in the experimental group were 10 to 12 percent less likely to have a report of child maltreatment.

Low-income families also experience intimate partner violence at higher rates than other families. For example, women living in households with annual incomes of less than $7,500 are seven times more likely to be victims of domestic violence than those with an income of at least $75,000 (Macomber, 2006). Furthermore, the rate of hospitalizations for domestic assault while pregnant is three times as high for mothers who are on

Medicaid than for those on private insurance (Aizer, 2011). These women are more likely to have poor birth outcomes such as low birth weight infants, fetal death, and increased infant mortality (Aizer, 2011). On average, infants born to mothers hospitalized for assault weigh 163 grams less at birth (Aizer, 2011). The lower birth weights of these infants are likely to lead to poorer heath, lower academic achievement, and reduced income in adulthood, thereby contributing to intergenerational poverty (Almond and Currie, 2011).

The causal relationships between poverty, child maltreatment, and intimate partner violence have been less clear. It is unknown if the higher rates of child maltreatment and domestic violence in low-income families reflect the increased stress of material and economic hardship, preexisting conditions (such as mental illness and substance abuse), the increased contact with mandatory reporters of child maltreatment, or attribution bias among mandatory reporters. For example, the material and economic hardships associated with poverty may lead to increased parenting stress and impact parents’ mental health and family relationships, which, in turn, may lead to violence or maltreatment of the child (for more on this process, see the section on mental health, below). Alternatively, conditions that can lead to poverty and to child maltreatment, such as parental mental health problems and substance abuse, are causing correlations between poverty and child maltreatment that are not causal. Children in low-income families are more likely to have a parent with poor mental health compared with children in higher income families (Macomber, 2006). As is often the case with a physical health problem, mental health problems can detract from the care and attention that a parent can provide their child. And finally, it is possible that the disparities in rates of child maltreatment and intimate partner violence between poor and higher-income families are a product of reporting bias. Poor families are more likely to come into contact with mandated reporters of child maltreatment through their participation in government systems, such as welfare. This increased visibility may lead to greater attention to problems in these families than in families with higher incomes, who are less likely to use welfare and other assistance programs.

MATERIAL HARDSHIP

Material hardship—broadly defined as the inability to meet basic needs—is associated with poverty. Many, but not all, of the negative impacts of poverty on child outcomes may be mediated through those material hardships. Measures of material hardship use indicators of consumption and physical living conditions that are directly related to whether families can meet basic needs, starting with needs for physiological functioning and survival (Ouellette et al., 2004). Therefore, measures of shelter

(housing, utilities), medical care, food security, and ability to pay for essential expenses are always included when researchers measure material hardship. Sometimes durable goods such as refrigerators, and neighborhood characteristics are also included (Ouellette et al., 2004).

The eighth wave of the 1996 panel of the Survey of Income and Program Participation (SIPP), a household survey conducted by the U.S. Census Bureau in 1998, included questions on material hardships and provided researchers the ability to look at the relationship of level of income poverty, as well as other characteristics of poverty (such depth and persistence) with various categories of material hardship. Across the board, all individual measures of material hardship were significantly associated with poverty, level of poverty, and measures of persistence and multiple “spells” of poverty (Iceland and Bauman, 2007; Figure D3-2; Table D3-2). The deeper the poverty, the more time in poverty, and the more “spells” of poverty, the stronger the relationship. Particularly strong relationships were seen for food insecurity, difficulty meeting basic needs (including paying for housing and other expenses, as well as medical care), and lack of durable goods (Iceland and Bauman, 2007). Even short spells of poverty seem to put families at greater risk for material hardships

TABLE D3-2 Material Hardship

compared with families who experienced no poverty. Hardships are also likely to coexist for poor families, with 71 percent of families below 50 percent of the federal poverty level (FPL) and 54 percent of families between 50 percent and 99 percent of FPL having two or more material hardships, and 38 percent of those below 50 percent of FPL and 23

percent of those between 50 percent and 99 percent of FPL having four or more hardships (Mayer and Jencks, 1989).

Poverty is significantly correlated with material hardship, which is an important mediator of poverty’s impact on childhood outcomes, at least in early childhood. For example, Gershoff and colleagues (2007) used a nationally representative sample of children developed by the U.S. Department of Education to look at the relationship of income poverty with childhood cognitive skills and social-emotional competence at kindergarten. They also examined the mediating effects of material hardship as well as those of parenting stress, positive parenting, and parental investment in their children. All of these mediators, including material hardship, were partial, meaning that there were still direct effects of poverty on child outcomes. However, material hardship was highly correlated with parenting stress, which negatively impacted parenting behaviors leading to worse outcomes in child social emotional competence. In fact, material hardship explained most of the impact of family income on social-emotional outcomes. Family income had its strongest direct effects on parent investment in their children, which in turn impacted child cognitive skills very strongly (Gershoff et al., 2007).

PHYSICAL HEALTH

Poor children face significant threats to their physical health compared with nonpoor children, starting at birth. They are more likely to be born at a low birth weight and to die during their first year of life; to experience an injury or poisoning requiring medical attention; to have elevated levels of lead in their blood; to experience a chronic disease such as asthma and obesity; and to experience food insecurity (Brooks-Gunn and Duncan, 1997; Chaudry and Wimer, 2016; Moore et al., 2009) (Table D3-3).

Researchers have documented a “gradient” between income and health status, both in children and adults: those with higher income have better health and in the case of adults, longevity (Case, Lubotsky, and Paxson, 2002). In adulthood, the direction of this gradient is difficult to determine. Does poor health lead to low income or does low income lead to poor health? In children, the direction of causality is clearer since children do not contribute to family income in the United States. Researchers have documented an inverse relationship between children’s health and family income, which becomes more negative as the child gets older (Case, Lubotsky, and Paxson, 2002). Even when controlling for parents’ education, the gradients remain strong, so that a doubling of family income increases the likelihood that the child is in very good or excellent health by 4 to 7 percent depending on the age of the child. The U.S. gradient is largest in those children with the most severe chronic conditions (Case, Lubotsky, and Paxson, 2002).

TABLE D3-3 Physical Health

Family income’s relationship to children’s health is in part due to its protection of children’s health on arrival of chronic diseases. In the United States, researchers using the Medical Expenditure Panel Survey (MEPS) as well as the Panel Study of Income Dynamics (PSID) have confirmed that poor children experience more new chronic health conditions than children in wealthier families, but also found that poorer families are less able to respond effectively to these chronic conditions (Case, Lubotsky, and Paxson, 2002). Parental income buffers children from the impacts of chronic diseases, and for almost every chronic condition, children from wealthier families experience better health (Case, Lubotsky, and Paxson, 2002). For example, poor children with asthma in the United States were almost 12 percent more likely to be in poor health 5 years later, whereas children with asthma from a higher-income family were only 4 percent more likely to be in poor health in that same time period (Condliffe and Link, 2008).

Studies from peer English-speaking nations with universal health care have documented a similar relationship between family income and child health, although the magnitude of the relationship is not as pronounced. For example, Currie and Stabile (2003), using a panel of Canadian children, found a flatter gradient between income and child health, although it also steepens as children get older. The data indicate that poor children are subject to more health shocks due to chronic diseases as they get older, which explains the steepening gradient with child age in Canada (Currie and Stabile, 2003). A study of Australian children also confirmed the existence of an income-child health gradient that is flatter than the U.S. gradient (Khanam, Nghiem, and Connelly, 2009). These studies from countries with universal health care indicate that availability of health insurance is not sufficient to eliminate the income gradient in child health, although it appears to reduce it. A more general look at the relationship between income and child health is provided in Institute of Medicine (2013, pp. 2-3).

Children with poorer health also spend less time in school and have fewer years of education, especially if they live in families with lower income (Currie, 2009). Poorer children, due to the income-child health gradient, enter adulthood with poorer health and lower educational attainment, likely leading to lower adult earnings (Case, Lubotsky, and Paxson, 2002). Thus, at least part of the intergenerational transmission of poverty may be due to the impact of family income on children’s health.

FETAL HEALTH AND HEALTH AT BIRTH

There is extensive research indicating that maternal poverty and disadvantage during the prenatal period has an especially significant impact on infant health, one that lasts and affects long-term outcomes (Table D3-4). We know that there are large inequalities in infant health at birth (which can be crudely measured by the incidence of low birth weight) and that these inequalities are associated with socioeconomic factors such as race, maternal education, marital status and income. Defining maternal disadvantage in this way, the incidence of low birth weight is more than three times higher among disadvantaged mothers compared with highly advantaged mothers (Aizer and Currie, 2014).

Maternal health behaviors during pregnancy can impact birth outcomes, and health behaviors during pregnancy are better among mothers with higher incomes. For example, smoking during pregnancy—which has been shown to increase risk of low birth weight and premature birth—is 18 times more prevalent among poor mothers than those who are highly advantaged, and randomized controlled trials of smoking cessation interventions show that reductions in smoking during pregnancy are associated with higher birth weights and less prematurity. Sibling studies (where the mother smoked during one pregnancy and not the other) show similar impact on low birth weight (Aizer and Currie, 2014).

Harmful environmental factors also impact infant health at birth. Studies have consistently shown that poor women are more likely to live near various sources of pollution, such as Superfund hazardous waste sites, factories emitting toxic substances, and water districts with poor drinking-water quality (Currie, 2009). They are also less likely to be able to move to cleaner areas (Currie, 2009; Currie et al., 2013). Natural experiments that change the level of pollution due to policy changes have been shown to reduce the incidence of low birth weight by more than 10 percent (Currie and Walker, 2011).

Maternal health and nutritional status also impact fetal and infant health. Poor women are more likely to have preexisting obesity, diabetes, hypertension, and asthma when they become pregnant (Aizer and Currie, 2014). They are also more likely to be exposed and to be susceptible

TABLE D3-4 Fetal Health and Health at Birth

(through lack of immunization for example) to contagious diseases like influenza (Currie and Schwandt, 2013). Using data from flu epidemics researchers have shown that influenza during pregnancy has negative effects on infant birth weights, primarily for mothers who have other indicators of poor health. Disadvantaged mothers are also about one-half as likely to gain the recommended weight during pregnancy compared with more advantaged mothers and nutritional interventions during pregnancy have been shown to increase infant birth weight (Ludwig and Currie, 2010).

In summary, there are many reasons that poor women are at increased risk for infants born with lower birth weight, prematurity, and poorer health. Maternal behaviors such as smoking, increased exposure to pollution, and violence, and worse maternal health and nutrition all have negative impacts on infant health at birth. Some studies have also explored the long-term impacts of insults occurring prenatally on infants’ health,

child and adolescent academic achievement, and on adult earnings (Currie and Rossin-Slater, 2015). Finally, we know that nutritional programs such as the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) and Supplemental Nutritional Assistance Program (SNAP), home visiting programs, and poverty-reduction programs all have proven evidence and/or potential promise to ameliorate these negative outcomes.

BRAIN DEVELOPMENT

Poor children are more likely than other children to have developmental, behavioral, and academic problems, such as serious emotional or behavioral difficulties and learning disabilities (Bradbury et al., 2015; Chaudrey and Wimer, 2016; Heckman, 2006; Moore et al., 2009; Yoshikawa, Aber, and Beardslee, 2012; Table D3-5). It is possible that these outcomes reflect structural brain changes in poor children. Children in poor families have recently been documented to have reduced volumes in the cerebral cortex and hippocampus in areas that are key for school readiness and academic achievement (National Scientific Council on the Developing Child, 2011). These areas are associated with executive function, language development, and memory. Other studies have shown that parental income is related to the surface area of the frontal, temporal, and parietal cortex of children (Noble et al, 2015).

Children growing up in poverty are often exposed to “toxic stress,” which can lead to permanent changes in brain structure and function. Toxic stress is characterized by strong, prolonged activation of the body’s stress response systems (Shonkoff et al., 2012). Poverty causes increased exposure to ACEs and material hardship, as described above, which may induce toxic stress. In addition, parental stress can increase the child’s experience of chronic unbuffered strong stress, which can be especially harmful in the period of rapid brain development that occurs in early childhood (National Council on the Developing Child, 2014; Shonkoff et al., 2012). Toxic stress activates the hormonal systems that respond to stress, especially the hypothalamic-pituitary-adrenocortical (HPA) system, causing sustained high levels of cortisol and corticotropin-releasing hormone, as well as of inflammatory cytokines (Shonkoff et al., 2012). These sustained high levels of stress hormones alter the size and neuronal architecture of the amygdala, hippocampus, and prefrontal cortex, which can lead to functional differences in learning, memory, executive functioning, and to an increased potential for long-term fear and anxiety (Shonkoff et al., 2012).

Elevated exposure to stress appears to modify the physiologic response to stress in ways that alter neuroendocrine activity and neural activity, and ultimately brain development and function, in ways that adversely affect the regulation of emotion and attention (Blair and Raver, 2016). Poor children

TABLE D3-5 Brain Development

experience more chronic stressors and generally manifest higher cortisol and other physiological stress markers than economically advantaged children (Blair et al., 2011; Chen, Cohen, and Miller, 2010; Evans, 2003; Evans and English, 2002; Evans et al., 2005). Exposure to chronic stressors is associated with more externalizing problems and a higher level of behavior problems in children (e.g., Coley, Lynch, and Kull, 2015). Given these connections, it is highly plausible that poverty affects children’s mental health through its effects on brain development (Blair and Raver, 2016).

There is increasing evidence that toxic stress in the prenatal period and early childhood can influence long-term child outcomes by chemically altering the structure and function of genes. These epigenetic changes alter gene expression without altering the genetic code itself, and effectively turn the gene “on” or “off.” These epigenetic changes may be permanent (and some

may be transferred intergenerationally). They may lead to impairments in learning abilities, and increased risk of mental illness, asthma, hypertension, heart disease and diabetes as adults. Not all epigenetic changes are harmful. Research has also shown that positive epigenetic changes in brain cells occur as cognition and memory develop, and repeated stimulation of these brain circuits through positive interactions with the environment and supportive adults (e.g., positive parenting), as well as good prenatal nutrition, enhance these important epigenetic changes (Institute of Medicine, 2000).

One biomarker for chronic stress is telomere length. Telomeres are DNA-protein complexes at the end of chromosomes. Telomeres shorten with age, and toxic stress is believed to accelerate shortening (Mitchell et al., 2014). A study of 9-year-old African American boys compared telomere length in children raised in disadvantaged environments and those in advantaged environments. Growing up in a disadvantaged environment is associated with a 19 percent shorter telomere, whereas a doubling of family income is associated with a 5 percent increase in telomere length (Mitchell et al., 2014).

In addition, severe psychosocial stress in pregnancy has been shown to be associated with shorter telomeres in young adult offspring, and young children exposed to violence have increased telomere shortening when tested at age 10 compared with telomere length when they were 5 (Price et al., 2013). A study of children growing up in poor neighborhoods with a high concentration of poverty, unemployment, and physical disorder documented the association of neighborhood level factors with reduced telomere length as well (Theall et al., 2013). Taken together, these studies provide evidence that chronic stress associated with socioeconomic disadvantage leads to accelerated telomere shortening, especially when experienced early in life. Since telomere length is a proxy for cellular aging, and is associated with diseases like diabetes, cancer, and heart disease, and with psychiatric conditions such as depression, this accelerated telomere shortening associated with poverty puts these children at risk for serious health and mental health problems as they move into adulthood.

MENTAL HEALTH

Numerous correlational studies based on community and national samples and various sources of information about children’s mental health (parent reports, teacher reports, self-reports) show that family-level economic resources are related to children’s mental health, although the associations are generally weaker than for children’s cognitive development and school achievement (e.g., Dubow and Ippolito, 1994; Brooks-Gunn and Duncan, 1997; Pagani, Boulerice, and Tremblay, 1997; Wadsworth et al., 2016; Table D3-6). Children from poor families are at increased risk of

TABLE D3-6 Mental Health

experiencing mental health problems, including depression, difficulties in peer relations, low self-esteem, antisocial behavior, and drug use compared with children in more socioeconomically advantaged families (Bolger et al., 1995; Currie and Lin, 2007; Evans and English, 2002; Gershoff et al., 2007; Goodman, 1999; Goosby, 2007; Strohschein, 2005; Wadsworth et al., 2005; Yoshikawa, Aber, and Beardslee, 2012). Generally, family poverty and other family-level economic indicators are more strongly related to children’s externalizing symptoms (e.g., disobedience, fighting, difficulty

getting along with others, impulsivity) than internalizing symptoms (e.g., anxiety, sadness-depression). Both externalizing symptoms and internalizing symptoms become more prevalent the longer children have been living in poverty (Bolger et al., 1995; Duncan et al., 1994; Goosby, 2007; Hanson, McLanahan, and Thomson, 1997; Korenman, Miller, and Sjaastad, 1995; McLeod and Shanahan, 1993; Pagani, Boulerice, and Tremblay, 1997).

In addition to differences in their income, parents and families who are poor differ in a variety of measured and unmeasured ways. Thus, the links between family income and children’s mental health may reflect these unmeasured differences rather than income differences (Morris and Gennetian, 2003). In virtually all of the studies reviewed here, investigators take some of these differences into account by including control variables (e.g., family structure, maternal education, ethnicity, maternal age at first birth) in their model estimates, which typically reduces the association between income and child mental health. Such studies provide better estimates of the effects of poverty on children than studies without such controls. Nonetheless, these correlational studies are insufficient to establish a causal link between poverty and children’s mental health.

Children’s mental health is related not only to family-level economic indicators, but to area-level economic conditions. These area-level studies focus on economic stressors outside the family, and may help to establish a causal relationship between economic conditions and mental health. However, they generally lack the data necessary to explore processes that account for these associations. In Currie and Tekin’s (2015) study based on administrative hospital data from four states between 2005 and 2009, increases in home foreclosures within zip codes were associated with increases in hospital and emergency room visits for mental health problems among all ages, including children and adolescents. Gassman-Pines and colleagues’ (2014) analysis of data between 1997 and 2009 indicated that statewide job losses due to mass closings and mass layoffs were associated with an increase in the following year in girls’ probability of suicidal ideation and suicide plans, and in non-Hispanic Black adolescents’ probability of suicide attempts. Research also links increases in state-level unemployment rates to greater prevalence of mental health problems among children, controlling for parental mental health (Golberstein, Gonzales, and Meara, 2016).

Correlational studies that link family income and child mental health have produced evidence suggestive of several mediating mechanisms. Most attention has been given to family-based processes that involve material deprivation, parents’ psychological distress, and parenting behavior—processes encapsulated in what is widely known as the “family stress model” (Elder, 1974; Elder and Conger, 1994; for a review of these studies, see Conger, Conger, and Martin, 2010). This model hypothesizes that economic hardship induces strain and pressure in parents. The strain

associated with the daily hassles of making ends meet in turn takes a toll on parents’ mental health, increases interparental conflict and discord, and ultimately interferes with positive parenting, which in turn undermines the child’s mental health.

As examples of empirical support for this model, Gershoff and colleagues’ (2007) investigation based on a representative national sample of kindergarteners showed that higher family income predicted decreased material hardship (e.g., less food insecurity, residential instability, and inadequacy of medical care) and decreased parent stress (i.e., lower levels of marital conflict, parenting stress, depressive symptomatology), which in turn predicted increased positive parenting (i.e., more warmth and cognitive stimulation, less physical punishment) and reduced child problem behaviors. In their ethnically diverse, low-income sample, Mistry and colleagues (2002) found that lower income and increased perceptions of economic pressure affected parenting behavior through an adverse impact on parents’ psychological well-being. Distressed parents reported feeling less effective and capable in disciplinary interactions with their child and were observed to be less affectionate in parent-child interactions. In turn, less than optimal parenting predicted lower teacher ratings of children’s positive social behavior and higher ratings of behavior problems.

Other research into the links between poverty and children’s mental health point to children’s perceptions of family economic difficulties (Dashiff et al., 2009) and their cumulative exposure to stressors (e.g., psychosocial stressors such as family turmoil and community violence, and physical stressors such as substandard housing, high levels of noise, and crowding) (Evans and English, 2002).

Mental health problems in childhood and adolescence, especially externalizing behavior problems, warrant efforts aimed at prevention or early treatment because of their high costs to individuals, families (e.g., costs of treatments), and society (e.g., higher levels of delinquent behavior, crime, and addiction into adulthood) (Golberstein, Gonzales, and Meara, 2016). Their consequences for outcomes in adulthood include lower educational attainment, higher rates of unemployment, and reduced earnings (for a review of these studies, see Currie, 2009). Mental health problems in childhood and adolescence tend to foreshadow mental health problems in adulthood, as data indicate that roughly one-half of all lifetime cases of mental health disorders start in childhood or adolescence (Kessler et al., 2005). Children’s mental health appears to play a significant role in the intergenerational transmission of poverty (Currie, 2009), given evidence linking poverty and low socioeconomic status to parents’ mental health (Lorant et al., 2003, 2007) and children’s mental health and the association between child mental health, future education, and labor market outcomes (Currie, 2009).

EDUCATIONAL ATTAINMENT

Innumerable studies have documented a gap in the average achievement levels of students from low-income families relative to high-income families (Table D3-7). This gap appears in virtually all measures of achievement including grades, standardized test scores, high school graduate rates, college attendance, and college graduation rates (Bradbury et al., 2015; Chaudry and Wimer, 2016; Haveman and Wolfe, 1995; Heckman, 2006; Moore et al., 2009; Yoshikawa, Aber, and Beardslee, 2012).

Alarmingly, Reardon (2013) documents the fact that this income-achievement gap is not stable, but has been increasing steadily over the past 50 years. He shows that the gap in achievement between a child at the 90th percentile of the family income distribution and a child at the 10th percentile of the family income distribution is almost twice as big as the gap in achievement between white and African-American students—while 50 years ago the situation was reversed.

This gap in achievement is present when children enter kindergarten and stays relatively stable over time, indicating that the gap has its origins in the preschool period. Indeed Hair et al. (2015) suggest that differences in brain volume in the frontal and temporal cortex can explain 15–20 percent of the difference in scores between poor and nonpoor children, suggesting that intervention during periods of peak brain development in early childhood may be key.

RISKY BEHAVIORS, CRIME, AND DELINQUENCY

Poverty is linked to adolescent delinquency independent of other familial factors such as maternal education, family configuration status, and earlier childhood behavior patterns (Pagani et al., 1999; Table D3-8). In general, this link is stronger when the dependent variable is serious delinquency (Bjerk, 2007; Farnworth et al., 1994; Jarjoura, Triplett, and Brinker, 2002), if youth live in persistent poverty (Duncan et al., 1994; Farnworth et al., 1994; Jarjoura, Triplett, and Brinker, 2002), and if adolescents experienced poverty during early childhood or during adolescence (Brooks-Gunn and Duncan, 1997; Duncan et al, 1998; Jarjoura, Triplett, and Brinker, 2002).

Bjerk’s (2007) systematic analyses suggest that the association between income and delinquency can be obscured because of its nonlinearity and because of error in the measurement of household economic resources. Dividing household income into quintiles, he found a negative relationship between household income and participation in serious crime (e.g., assault, stealing with a weapon or use of force, selling “hard drugs”) only when comparing adolescents at the lowest end of the income distribution

TABLE D3-7 Educational Attainment

TABLE D3-8 Risky Behaviors, Crime, and Delinquency

with those at the highest end of the income distribution. In addition, the relationship was much stronger if the measure of family economic well-being included other family economic indicators in addition to income (e.g., savings, debts owed, inheritances). Household income and participation in minor crime (e.g., stealing without force or threat of force, property destruction) were unrelated.

Family poverty in middle childhood appears to be less important as an antecedent to serious delinquency than poverty during early childhood or adolescence (Jarjoura, Triplett, and Brinker, 2002). Several studies show that much of the increased risk of delinquency associated with poverty is mediated through negative parenting and family conflict (e.g., harshness, inconsistency, low monitoring and involvement) (Conger et al., 1994, 1995;

Rutter, Giller, and Hagell, 1998; Sampson and Laub, 1994), though this mediation is not found in some studies (e.g., Pagani et al., 1999).

Several studies have found that official intervention (e.g., being arrested or convicted) during adolescence is negatively associated with employment in early adulthood (Aizer and Doyle, 2015). Moreover, analyses of panel data suggest that official intervention increases involvement in crime in early adulthood by adversely affecting educational attainment and employment. Declines in educational progress following official intervention may be triggered by stigma and exclusion from school. Increases in crime in response to official intervention appear to be especially pronounced among African American males and males who come from poor family backgrounds (Bernburg and Krohn, 2003).

Delinquency is one of a group of adolescent risky behaviors that are correlated and tend to co-occur (Gruber, 2001; Jessor, Turbin, and Costa, 1998). This co-occurrence has stimulated research on domains and profiles of adolescent risky behavior and their antecedents and outcomes in early adulthood (e.g., Bartlett, Holditch-Davis, and Belyea, 2005; Hair et al., 2009; Zweig, Lindberg, and McGinley, 2001). Blum et al. (2000) focused specifically on health-compromising behaviors among 7th–12th graders (e.g., alcohol use, cigarette smoking, suicidal thoughts or attempts, weapon-related violence, sexual intercourse) in the National Longitudinal Study of Adolescent Health (Add Health). Lower family income was associated with less alcohol use (among 9th–12th graders), but higher weapon-related violence, and greater likelihood of sexual intercourse, controlling for demographic variables related to family income. However, family income, family structure, and race/ethnicity, taken together, explained very little of the variance in these outcomes.

Using data from over 12,500 adolescents from Add Health, Bartlett, Holditch-Davis, and Belyea (2005) identified three clusters of youth (mean age 15.8 years) based on 14 problem behaviors, including delinquency (e.g., stealing, weapon use, property damage), alcohol and marijuana use, sexual behavior (e.g., multiple sex partners, sex without birth control) and truancy. Youth in the “normal” cluster reported few, if any problem behaviors; those in the “deviant” cluster had high means for most of the behaviors and the largest number of problem behaviors reflecting deviant, conduct-type problems like selling drugs and weapon use; youth in the “problem” cluster had higher means than those in the “normal” cluster for most problems, but lower means than youth in the “deviant” cluster. Family SES was a significant predictor of cluster membership, with adolescents in the “normal” cluster having significantly higher SES than adolescents in the “deviant” and “problem” clusters. Other research distinguishing profiles of risky behavior among adolescents (based on delinquency, smoking, drug use, drinking, sexual behavior, etc.) finds worse adult outcomes (e.g., increased

probability of arrest, not having a high school diploma or GED, not being employed) among those in higher-risk groups, compared with those in lower-risk groups (Hair et al., 2009).

Early timing of events beyond sexual debut has been conceptualized as a marker of risky behavior (e.g., parenthood) because these early transitions often create demands and conditions that place individuals at increased risk for low education and enduring occupational, financial, and attendant stressors that compromise psychological health (McLoyd, Purtell, and Hardaway, 2015). Research confirms that family poverty and low socioeconomic status increase the probability of early timing of transition events (e.g., sexual intercourse by age 14; pregnancy, parenthood, cohabitation, marriage, or leaving the parental home by age 17) partly by increasing the occurrence of negative life events experienced by parents or the family (e.g., getting laid off, taking wage cuts, moving to worse residences or neighborhoods). In turn, early timing of transition events has been found to predict significant growth in depressive symptoms in early adulthood (Wickrama, Merten, and Elder, 2005; Wickrama et al., 2008).

THE TIMING OF POVERTY

The timing of poverty is important. Early childhood is a period when brain development is rapid, and children are very sensitive to the impacts of family poverty (Blair and Raver, 2016). Language development diverges for poor and nonpoor children almost as soon as expressive language emerges at 15 or 16 months of age, and by 3 years of age poor children are markedly behind in their language acquisition (Hart and Risley, 1995). One recent study indicated that differences in language development between poor and nonpoor children can be seen as early as 7 months of age (Betancourt, Brodsky, and Hurt, 2015). Another study looking at EEG³⁷ patterns found decreased electrical activity in the frontal cortex, the part of the brain that controls executive function, in poor 6- to 9-month-olds compared with those living in families with higher income (Tomalski et al., 2013). The impacts of poverty experienced early in childhood last into adulthood. Poverty in early childhood has been shown to be a very significant negative predictor of academic performance in school in middle childhood and beyond. The reading and math skills of children experiencing poverty in early life diverge over time from the skills of more advantaged children during the school years (Votruba-Drzal, 2006). Children who are poor from birth to age 2 are 30 percent less likely to graduate from high school and are three times more likely to have a teen premarital pregnancy (Ratcliffe and

___________________

³⁷ An EEG, or electroencephalogram, is a neurological test that detects abnormalities in the brain’s electrical activity.

McKernan, 2012). These outcomes are likely to lead to less employment as adults and to intergenerational poverty. And poverty in early childhood is likely to persist. Nearly one-half of children born to poor parents remain poor for one-half or more of their childhoods (Ratcliffe and McKernan, 2012).

Nevertheless, poverty experienced later in childhood is also associated with negative outcomes in adolescence and adulthood, perhaps in part because the length of time a child spends in poverty is also important. Experiencing persistent poverty (for one-half or more of childhood years) is associated with not graduating from high school, having teen nonmarital births and nonmarital births as adults for females, and with higher arrest rates by young adulthood for males (Ratcliffe and McKernan, 2012). As with poverty experienced in early childhood, these outcomes lead to lower earnings as adults and an increased likelihood of long-term adult poverty and poorer adult health (Chaudry and Wimer, 2016; Ratcliffe and McKernan, 2012). The longer the duration of poverty, the more likely the child will have these negative outcomes.

SEVERITY OF POVERTY

In 2015, 2.9 percent of children in the United States, or 2.1 million children,³⁸ lived in deep poverty, that is, had family income less than 50 percent of the Supplemental Poverty Measure. Families living in deep poverty experience even greater material hardship and parenting stress than those who are poor but living between 50 percent and 99 percent of the federal poverty level (Mayer and Jencks, 1989). Young children growing up in deep poverty have higher rates of obesity, and three times the rate of elevated blood lead levels compared with other poor children (Ekono, Jiang, and Smith, 2016). One study showed that children in deep poverty had scores 6 to 13 points lower on standardized tests of IQ, verbal ability, and achievement compared with nonpoor children. Scores for children living in poverty but above deep poverty were also lower than those who were nonpoor, but the differences were not as large (Brooks-Gunn and Duncan, 1997).

Children in deep poverty also are more likely to grow up in families with significant additional risk factors. Compared with poor children not in deep poverty, they are more likely to have parents reporting poor or fair health and mental health, more parental stress, less social support, and living in unsafe neighborhoods (Ekono, Jiang, and Smith, 2016). These factors predict poor health and development outcomes in children. The combination of deep poverty and family adversity is particularly toxic (Ekono, Jiang, and Smith, 2016).

___________________

³⁸ Per TRIM3 weighted estimates commissioned for the committee.

REFERENCES

Condliffe, S., and Link, C.R. (2008). The relationship between economic status and child health: Evidence from the United States. American Economic Review, 98(4), 1605–1618.

Conger, R.D., Conger, K.J., and Martin, M.J. (2010). Socioeconomic status, family processes, and individual development. Journal of Marriage and Family, 72(3), 685–704.

Conger, R.D., Ge, X., Elder, G.H., Jr., Lorenz, F.O., and Simons, R.L. (1994). Economic stress, coercive family process, and developmental problems of adolescents. Child Development, 65(2 Spec No), 541–561.

Conger, R.D., Patterson, G.R., and Ge, X. (1995). It takes two to replicate: A mediational model for the impact of parents’ stress on adolescent adjustment. Child Development, 66(1), 80–97.

Currie, J. (2011). Inequality at birth: Some causes and consequences. American Economic Review, 101(3), 1–22.

Currie, J. (2009). Healthy, wealthy, and wise: Socioeconomic status, poor health in childhood, and human capital development. Journal of Economic Literature, 47(1), 87–122.

Currie, J., and Lin, W. (2007). Chipping away at health: More on the relationship between income and child health. Health Affairs (Millwood), 26(2), 331–344.

Currie, J., and Rossin-Slater, M. (2015). Early-life origins of life-cycle well-being: Research and policy implications. Journal of Policy Analysis and Management, 34(1), 208–242.

Currie, J., and Schwandt, H. (2013). Within-mother analysis of seasonal patterns in health at birth. Proceedings of the National Academy of Sciences, 110(30), 12265–12270.

Currie, J., and Stabile, M. (2003). Socioeconomic status and child health: Why is the relationship stronger for older children? American Economic Review, 93(5), 1813–1823.

Currie, J., and Tekin, E. (2015). Is there a link between foreclosure and health? American Economic Journal: Economic Policy, 7(1), 63–94.

Currie, J., and Walker, R. (2011). Traffic congestion and infant health: Evidence from E-ZPass. American Economic Journal-Applied Economics, 3(1), 65–90.

Currie, J., Zivin, J.G., Meckel, K., Neidell, M., and Schlenker, W. (2013). Something in the water: Contaminated drinking water and infant health. Canadian Journal of Economics, 46(3). https://doi.org/10.1111/caje.12039.

Dashiff, C., DiMicco, W., Myers, B., and Sheppard, K. (2009). Poverty and adolescent mental health. Journal of Child and Adolescent Psychiatric Nursing, 22(1), 23–32.

Dubow, E.F., and Ippolito, M.F. (1994). Effects of poverty and quality of the home-environment on changes in the academic and behavioral-adjustment of elementary school-age-children. Journal of Clinical Child Psychology, 23(4), 401–412.

Duncan, G., Brooks-Gunn, J., and Klebanov, P. (1994). Economic deprivation and early childhood development. Child Development, 65, 296–318.

Duncan, G.J., Yeung, W.J., Brooks-Gunn, J., and Smith, J.R. (1998). How much does childhood poverty affect the life chances of children? American Sociological Review, 63(3), 406.

Ekono, M.M., Jiang, Y., and Smith, S. (2016). Young Children in Deep Poverty. New York: National Center for Children in Poverty, Columbia University. Available: https://files.eric.ed.gov/fulltext/ED578993.pdf.

Elder, G.H. (1974). Children of the Great Depression. Chicago: University of Chicago Press.

Elder Jr., G.H., and Conger, R.D. (1994). Families in Troubled Times: Adapting to Change in Rural America. Social Institutions and Social Change. Piscataway, NJ: Aldine Transaction.

Evans, G.W. (2003). A multimethodological analysis of cumulative risk and allostatic load among rural children. Developmental Psychology, 39(5), 924–933.

Evans, G.W., and English, K. (2002). The environment of poverty: Multiple stressor exposure, psychophysiological stress, and socioemotional adjustment. Child Development, 73(4), 1238–1248.

Evans, G.W., Gonnella, C., Marcynyszyn, L.A., Gentile, L., and Salpekar, N. (2005). The role of chaos in poverty and children’s socioemotional adjustment. Psychological Science, 16(7), 560–565.

Farnworth, M., Thornberry, T., Krohn, M., and Lizotte, A. (1994). Measurement in the study of class and delinquency: Integrating theory and research. Journal of Research in Crime and Delinquency, 31(1), 32–61.

Gassman-Pines, A., Ananat, E.O., and Gibson-Davis, C.M. (2014). Effects of statewide job losses on adolescent suicide-related behaviors. American Journal of Public Health, 104(10), 1964–1970.

Gershoff, E.T., Aber, J.L., Raver, C.C., and Lennon, M.C. (2007). Income is not enough: Incorporating material hardship into models of income associations with parenting and child development. Child Development, 78(1), 70–95.

Golberstein, E., Gonzales, G., and Meara, E. (2016). Economic Conditions and Children’s Mental Health. Cambridge, MA: National Bureau of Economic Research.

Goodman, E. (1999). The role of socioeconomic status gradients in explaining differences in US adolescents’ health. American Journal of Public Health, 89(10), 1522–1528.

Goosby, B.J. (2007). Poverty duration, maternal psychological resources, and adolescent socioemotional outcomes. Journal of Family Issues, 28(8), 1113–1134.

Gruber, J. (2001). Risky Behavior Among Youths. Chicago: University of Chicago Press.

Hair, E.C., Park, M.J., Ling, T.J., and Moore, K.A. (2009). Risky behaviors in late adolescence: Co-occurrence, predictors, and consequences. Journal of Adolescent Health, 45(3), 253–261.

Hair, N.L., Hanson, J.L., Wolfe, B.L., and Pollak, S.D. (2015). Association of child poverty, brain development, and academic achievement. JAMA Pediatrics, 169(9), 822–829.

Hanson, T., McLanahan, S., and Thomson, E. (1997). Economic resources, parental practices, and children’s well-being. In G. Duncan and J. Brooks-Gunn (Eds.), Consequences of Growing Up Poor. New York: Russell Sage Foundation.

Hart, B., and Risley, T.R. (1995). Meaningful Differences in the Everyday Experience of Young American Children. Baltimore, MD: Paul H Brookes Publishing.

Haveman, R., and Wolfe, B. (1995). The determinants of children’s attainments: A review of methods and findings. Journal of Economic Literature, 33(4), 1829–1878.

Heckman, J.J. (2006). Skill formation and the economics of investing in disadvantaged children. Science, 312(5782), 1900–1902.

Iceland, J., and Bauman, K.J. (2007). Income poverty and material hardship: How strong is the association? The Journal of Socio-Economics, 36(3), 376–396.

Institute of Medicine. (2013). U.S. Health in International Perspective—Shorter Lives, Poorer Health Report Brief, January. Washington, DC: The National Academies Press. Available: http://www.nationalacademies.org/hmd/~/media/Files/Report%20Files/2013/US-Health-International-Perspective/USHealth_Intl_PerspectiveRB.pdf.

Institute of Medicine (2000). From Neurons to Neighborhoods: The Science of Early Childhood Development. Washington, DC: National Academy Press.

Jarjoura, G.R., Triplett, R.A., and Brinker, G.P. (2002). Growing up poor: Examining the link between persistent childhood poverty and delinquency. Journal of Quantitative Criminology, 18(2), 159–187.

Jessor, R., Turbin, M.S., and Costa, F.M. (1998). Risk and protection in successful outcomes among disadvantaged adolescents. Applied Developmental Science, 2(4), 194–208.

Kessler, R.C., Berglund, P., Demler, O., Jin, R., Merikangas, K.R., and Walters, E.E. (2005). Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the national comorbidity survey replication. Archives of General Psychiatry, 62(6), 593–602.

Khanam, R., Nghiem, H.S., and Connelly, L.B. (2009). Child health and the income gradient: Evidence from Australia. Journal of Health Economics, 28(4), 805–817.

Korenman, S., Miller, J.E., and Sjaastad, J.E. (1995). Long-term poverty and child development in the United States: Results from the NLSY. Children and Youth Services Review, 17(1), 127–155.

Lorant, V., Deliege, D., Eaton, W., Robert, A., Philippot, P., and Ansseau, M. (2003). Socioeconomic inequalities in depression: A meta-analysis. American Journal of Epidemiology, 157(2), 98–112.

Lorant, V., Croux, C., Weich, S., Deliege, D., Mackenbach, J., and Ansseau, M. (2007). Depression and socio-economic risk factors: 7-year longitudinal population study. British Journal of Psychiatry, 190, 293–298.

Ludwig, D.S., and Currie, J. (2010). The association between pregnancy weight gain and birthweight: A within-family comparison. Lancet, 376(9745), 984–990.

Macomber, J. (2006). An Overview of Selected Data on Children in Vulnerable Families. Washington, DC: Urban Institute.

Mayer, S.E., and Jencks, C. (1989). Poverty and the distribution of material hardship. Journal of Human Resources, 24(1), 88–114.

McLeod, J.D., and Shanahan, M.J. (1993). Poverty, parenting, and children’s mental health. American Sociological Review, 351–366.

McLoyd, V.C., Purtell, K.M., and Hardaway, C.R. (2015). Race, class, and ethnicity in young adulthood. In Handbook of Child Psychology and Developmental Science. Hoboken, NJ: John Wiley and Sons.

Mistry, R.S., Vandewater, E.A., Huston, A.C., and McLoyd, V.C. (2002). Economic well-being and children’s social adjustment: The role of family process in an ethnically diverse low-income sample. Child Development, 73(3), 935–951.

Mitchell, C., Hobcraft, J., McLanahan, S.S., Siegel, S.R., Berg, A., Brooks-Gunn, J., Garfinkel, I., and Notterman, D. (2014). Social disadvantage, genetic sensitivity, and children’s telomere length. Proceedings of the National Academy of Sciences, 111(16), 5944–5949.

Moore, K.A., Redd, Z., Burkhauser, M., Mbwana, K., and Collins, A. (2009). Children in Poverty: Trends, Consequences, and Policy Options. Washington, DC: Child Trends.

Morris, P.A., and Gennetian, L.A. (2003). Identifying the effects of income on children’s development using experimental data. Journal of Marriage and Family, 65(3), 716–729.

National Scientific Council on the Developing Child (2011). Building the Brain’s “Air Traffic Control” System: How Early Experiences Shape the Development of Executive Function. Cambridge, MA: Harvard University.

National Scientific Council on the Developing Child (2014). Excessive Stress Disrupts the Architecture of the Developing Brain: Working Paper 3 Updated Edition. Cambridge, MA: Harvard University. Available: https://developingchild.harvard.edu/wp-content/uploads/2005/05/Stress_Disrupts_Architecture_Developing_Brain-1.pdf [August 30, 2018].

National Survey of Children’s Health 2011/12. (2017). Adverse Childhood Experiences. Data query from the Child and Adolescent Health Measurement Initiative, Data Resource Center for Child and Adolescent Health website. Retrieved September 28, 2018 from www.childhealthdata.org.

Noble, K.G., Houston, S.M., Brito, N.H., Bartsch, H., Kan, E., Kuperman, J.M., Akshoomoff, N., Amaral, D.G., Bloss, C.S., Libiger, O., Schork, N.J., Murray, S.S., Casey, B.J., Chang, L., Ernst, T.M., Frazier, J.A., Gruen, J.R., Kennedy, D.N., Van Zijl, P., Mostofsky, S., Kaufmann, W.E., Kenet, T., Dale, A.M., Jernigan, T.L., and Sowell, E.R. (2015). Family income, parental education and brain structure in children and adolescents. Nature Neuroscience, 18(5), 773–778.

Ouellette, T., Burstein, N., Long, D., and Beecroft, E. (2004). Measures of Material Hardship: Final Report. Washington, DC: US Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation.

Pagani, L., Boulerice, B., Tremblay, R. (1997). The influence of poverty on children’s classroom placement and behavior problems. In G.J. Duncan and J. Brooks-Gunn (Eds.), Consequences of Growing Up Poor (pp. 311–329). New York: Russell Sage Foundation.

Pagani, L., Boulerice, B., Vitaro, F., and Tremblay, R.E. (1999). Effects of poverty on academic failure and delinquency in boys: A change and process model approach. Journal of Child Psychology and Psychiatry, 40(8), 1209–1219.

Paxson, C., and Waldfogel, J. (2003). Welfare reforms, family resources, and child maltreatment. Journal of Policy Analysis and Management, 22(1), 85–113.

Price, L.H., Kao, H.T., Burgers, D.E., Carpenter, L.L., and Tyrka, A.R. (2013). Telomeres and early-life stress: An overview. Biological Psychiatry, 73(1), 15–23.

Ratcliffe, C.E., and McKernan, S.M. (2012). Child Poverty and Its Lasting Consequence. Washington, DC: Urban Institute. Available: https://www.urban.org/sites/default/files/publication/32756/412659-Child-Poverty-and-Its-Lasting-Consequence.PDF.

Reardon, S.F. (2013). The widening income achievement gap. Educational Leadership, 70(8), 10–16.

Rutter, M., Giller, H., and Hagell, A. (1998). Antisocial Behaviour by Young People. New York, NY: Cambridge University Press.

Sampson, R.J., and Laub, J.H. (1994). Urban poverty and the family context of delinquency: A new look at structure and process in a classic study. Child Development, 65(2 Spec No), 523–540.

Sedlak, A.J., Mettenburg, J., Basena, M., Petta, I., McPherson, K., Greene, A., and Li, S. (2010). Fourth National Incidence Study of Child Abuse and Neglect (NIS-4): Report to Congress: Executive Summary. Washington, DC: U.S. Department of Health and Human Services. Available: https://www.acf.hhs.gov/sites/default/files/opre/nis4_report_congress_full_pdf_jan2010.pdf [August 30, 2018].

Shonkoff, J.P., Garner, A.S., Committee on Psychosocial Aspects of Child and Family Health, Committee on Early Childhood, Adoption and Family Care, and Section on Developmental and Behavioral Pediatrics. (2012). The lifelong effects of early childhood adversity and toxic stress. Pediatrics, 129(1), e232–e246.

Strohschein, L. (2005). Household income histories and child mental health trajectories. Journal of Health and Social Behavior, 46(4), 359–375.

Theall, K.P., Brett, Z.H., Shirtcliff, E.A., Dunn, E.C., and Drury, S.S. (2013). Neighborhood disorder and telomeres: Connecting children’s exposure to community level stress and cellular response. Social Science and Medicine, 85, 50–58.

Tomalski, P., Moore, D.G., Ribeiro, H., Axelsson, E.L., Murphy, E., Karmiloff-Smith, A., Johnson, M.H., and Kushnerenko, E. (2013). Socioeconomic status and functional brain development—associations in early infancy. Developmental Science, 16(5), 676–687.

Votruba-Drzal, E. (2006). Economic disparities in middle childhood development: Does income matter? Developmental Psychology, 42(6), 1154–1167.

Wadsworth, M.E., Evans, G.W., Grant, K., Carter, J.S., and Duffy, S. (2016). Poverty and the development of psychopathology. In D. Cicchetti (Ed.), Developmental Psychopathology (pp. 136–179, 3rd edition). Hoboken, NJ: Wiley.

Wadsworth, M.E., Raviv, T., Compas, B.E., and Connor-Smith, J.K. (2005). Parent and adolescent responses to poverty related stress: Tests of mediated and moderated coping models. Journal of Child and Family Studies, 14(2), 283–298.

Wickrama, K.A.S., Merten, M.J., and Elder, G.H. Jr. (2005). Community influence on precocious transitions to adulthood: Racial differences and mental health consequences. Journal of Community Psychology, 33(6), 639–653.

Wickrama, K.A.S., Conger, R.D., Lorenz, F.O., and Jung, T. (2008). Family antecedents and consequences of trajectories of depressive symptoms from adolescence to young adulthood: A life course investigation. Journal of Health and Social Behavior, 49(4), 468–483.

Yoshikawa, H., Aber, J.L., and Beardslee, W.R. (2012). The effects of poverty on the mental, emotional, and behavioral health of children and youth: Implications for prevention. American Psychologist, 67(4), 272–284.

Zweig, J.M., Lindberg, L.D., and McGinley, K.A. (2001). Adolescent health risk profiles: The co-occurrence of health risks among females and males. Journal of Youth and Adolescence, 30(6), 707–728.

APPENDIX D, 4-1
DEFINITIONS PERTAINING TO CHAPTER 4 FROM THE ORGANISATION FOR ECONOMIC CO-OPERATION AND DEVELOPMENT (OECD)

Definition of OECD Family Benefits Public Spending³⁹

Figure 4-12 in Chapter 4 of this report presents data on public spending on families and children as a percent of the Gross Domestic Product (GDP). This is based on an indicator known as OECD Family Benefits Public Spending, which refers to public spending on family benefits, including financial support that is exclusively for families and children. Spending on health and housing also assists families, but not exclusively, and it is not included in this indicator. Broadly speaking, there are three types of public spending on family benefits:

1. Public spending on child-related cash transfers to families with children, including child benefits (or child allowances) that in some countries are income-tested; public income support payments for single-parent families; and income support issued during periods of parental leave.
2. Public spending on services (benefits in kind) for families with children, including direct financing and subsidizing of providers of child care and early education facilities; public child care support through earmarked payments to parents; public spending on assistance for young people and residential facilities; public spending on family services, including center-based facilities; and home help services for families in need.
3. Financial support for families provided through the tax system, including tax exemptions (e.g., income from child benefits that is not included in the tax base); child tax allowances (amounts for children that are deducted from gross income and are not included in taxable income); and child tax credits, amounts that are deducted from the tax liability.

This indicator can be broken down by cash benefits and benefits in kind and is measured in percentage of GDP.

___________________

³⁹ See https://data.oecd.org/socialexp/family-benefits-public-spending.htm.

APPENDIX D, 4-2
GOVERNMENT POLICIES AFFECTING CHILD POVERTY IN AUSTRALIA AND IRELAND

The Australian Family Tax Benefit, Parts A and B

The Australian government provides a family tax benefit to families with children based on specific eligibility criteria. This family benefit is intended to assist with the cost of raising children and consists of two parts, A and B. Overall eligibility criteria include these: (1) the parent(s) have a dependent child or full-time student under the age of 20 who does not receive a pension, payment, or other benefits; (2) the parent(s) are providing care for the child at least 35 percent of the time; and (3) the family meets a specific income test.⁴⁰ The income test for this program, and for Australian transfer programs in general, is designed to eliminate eligibility only for very-high-income families. Consequently, it is a near universal benefit and is classified by OECD as such. Part A is given for each child in families that meet the eligibility criteria, and Part B is intended to provide additional assistance to single parents, nonparental caregivers, and couples with one earner. In order to receive Part A, children must also meet immunization requirements.

According to a United Nations Children’s Fund (UNICEF) report on the impact of the worldwide financial/economic crisis on child well-being in 41 high-income countries (Fanjul, 2014), based on an anchored poverty line Australia experienced the third-best improvement over the 2008–2012 period, with its child poverty rate falling from just over 19 percent to 13 percent. During this 4-year period, Australia moved from having the 19th-lowest rate to the 7th-lowest rate of child poverty. The UNICEF report highlights that Australia had a multipronged approach, which included countercyclical policies to alleviate the effects of the economic downturn as well as stimulus packages that were targeted to low-income families with children.

However, there is concern that going forward, Australia’s outcomes may not be as positive. In 2009, the indexation of Family Tax Benefit Part A was made less generous, leading to a decline in the number of children benefiting from it, which fell from about 80 percent of all dependent children in 2009 to about 69 percent in 2012 (Whiteford, 2014).

___________________

⁴⁰ Australian Government Department of Human Services, https://www.humanservices.gov.au/individuals/services/centrelink/family-tax-benefit.

Ireland’s National Policy Framework for Children and Young People

In 2014, Ireland’s government established the goal of lifting 70,000 children out of poverty by the year 2020, part of its National Policy Framework for Children and Young People, also known as the Better Outcomes, Brighter Futures (BOBF) Framework (Ireland, Department of Employment Affairs and Social Protection, 2017). This is based on 2013 guidance from the European Commission that set out a three-pronged framework to address child poverty. These prongs were “access to adequate resources; access to affordable quality services; and children’s right to participate.” (Ireland, Department of Employment Affairs and Social Protection, 2017, p. 11).

With regard to adequate resources, Ireland provides a number of different income supports for families with children. The government offers a Child Benefit, which is a monthly payment payable to the parents or guardians of children under the age of 16, or under the age of 18 if the child is in full-time education, full-time training, or with a disability and cannot support themselves (Ireland, Department of Employment Affairs and Social Protection, 2018). Ireland also offers:

• Increases in some social welfare payments for each qualified child,
• Family Income Supplement,
• Back to School Clothing and Footwear Allowance,
• School Meals,
• Back to Work Family Dividend (allowing parents to combine welfare and work),
• Domiciliary Care Allowance (for children under age 16 with a severe disability),
• Fuel Allowance, and
• Exceptional Needs Payments (Ireland, Department of Employment Affairs and Social Protection, 2017).

The Irish government also has programs intended to make quality child care more affordable and accessible. A detailed report provided by Ireland’s Department of Employment Affairs and Social Protection (2017) describes in depth how the Irish government is tackling child poverty and reports that in 2015, the country saw its first reduction in the number of children in consistent poverty since 2008.

REFERENCES

Fanjul, G. (2014, September). Innocenti Report Card 12. Florence, Italy: United Nations Children’s Fund.

Ireland, Department of Employment Affairs and Social Protection. (2017, October). Better Outcomes, Brighter Futures: Whole of Government Approach to Tackling Child Poverty. Available: https://www.welfare.ie/en/downloads/WholeOfGovernmentApproachToTacklingChildPoverty_BOBF_Final.pdf [April 19, 2018].

Ireland, Department of Employment Affairs and Social Protection. (2018, April). Your Guide to Our Schemes and Services: Families and Children. Available: https://www.welfare.ie/en/pdf/DK6.pdf [April 19, 2018].

Isaacs, J. B., Lou, C., Hahn, H., Ovalle, J., and Steuerle, C.E. (2018). Kids’ Share 2018: Report on Federal Expenditures on Children Through 2017 and Future Projections. Washington, DC: Urban Institute. Available: https://www.urban.org/sites/default/files/publication/98725/kids_share_2018_0.pdf [August 30, 2018].

Nichols, A. (2013). Explaining Changes in Child Poverty Over the Past Four Decades. Washington, DC: Urban Institute. Available: https://www.urban.org/research/publication/ explaining-changes-child-poverty-over-past-four-decades [August 29, 2018].

Whiteford, P. (2014). Australia bucks child poverty trend but the future looks a lot bleaker. The Conversation, October. Available: https://theconversation.com/australia-bucks-child-poverty-trend-but-the-future-looks-a-lot-bleaker-33456 [November 8, 2017].

Wimer, C. (2017). Original Analyses Commissioned for the Committee on Building an Agenda to Reduce the Number of Children in Poverty by Half in 10 Years.

TABLE D4-2 Estimated Change in Child Poverty If Current Programs Were Eliminated

NOTE: ACTC = Additional Child and Dependent Care Tax Credit; EITC = Earned Income Tax Credit; Other = Veterans Benefits (non means-tested), State Temporary Disability Benefits, and Black Lung Miner Benefits; SNAP = Supplemental Nutrition Assistance Program; SSI = Supplemental Security Income; UC = Unemployment Compensation; WC = Worker’s Compensation.

^a Social Security includes Social Security Income (including Social Security Retirement, Social Security Disability, Social Security Survivors, and Railroad Retirement).

^b Other benefits include Temporary Assistance for Needy Families (TANF), Solely State-Funded Assistance, Other public assistance, means-tested veterans’ benefits, means-tested education assistance, Low Income Home Energy Assistance Program (LIHEAP), National School Lunch Program, and the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC).

SOURCE: Original analyses commissioned by the committee from TRIM3.

APPENDIX D, 5-1
ADJUSTING ESTIMATES OF POVERTY REDUCTION FOR BEHAVIORAL EFFECTS

As discussed at the beginning of Chapter 5, tax and transfer programs may change behaviors in ways that could magnify or moderate program impacts on poverty. Key behaviors involve labor market (employment and hours of work) and family structure choices (marriage and fertility). For example, evidence consistently suggests that the Earned Income Tax Credit (EITC) increases employment and earnings among single mothers (Nichols and Rothstein, 2016). This strong pro-work effect (coupled with little evidence of earnings reduction for those already in the labor market) strengthens the anti-poverty effect of the program over what it would be if family income only increased by the amount of the initial benefit. On the other hand, that benefits phase out at higher income levels in income-tested programs like the Supplemental Nutrition Assistance Program (SNAP), public housing, and the Supplemental Security Income (SSI) program can lead to reductions in employment and/or hours worked. Earning reductions weaken the anti-poverty effects of the programs over what they would be if family income only increased by the amount of the benefit.

A large volume of scholarly research on behavioral effects of policies over the last 40 years has shown that policies can and sometimes do affect employment and hours of work, although the magnitude of the impacts vary across studies and often appear only for some population groups. However, while transfer programs frequently affect the employment and hours of work of their recipients, a given program’s caseload is often too small to change the aggregate poverty rate very much, leading one recent review to conclude that, while there are significant behavioral side effects of many programs, they have little effect on the aggregate poverty impact of the safety net system (Ben-Shalom, Moffitt, and Scholz, 2012). Nevertheless, because behavioral effects on employment and hours of work can be nonnegligible if the caseload is large and if the impact on recipients is significant, the committee’s judgments regarding consensus estimates of these behavioral effects are incorporated in the poverty estimates reported in Chapters 5 and 6. Program-by-program details on our behavioral assumptions are provided in this appendix, with additional implementation details provided in Appendix F.

A smaller research literature attempts to estimate behavioral effects of programs and policies on family structure and childbearing. As described in Chapter 7, estimates from this research are much more tenuous and variable than those for the effects of programs and policies on labor market behavior. More often than not, no statistically significant responses are found. As a result, the committee did not simulate behavioral responses on family structure and childbearing. We refer to this evidence selectively below.