Abstract
Advances in poverty measurement have opened new opportunities for investigating differences in poverty among racial and ethnic groups. Some researchers have attributed differences in poverty to differences in group characteristics, such as marital status or educational attainment, whereas others have focused on labor market differences or to differential benefits from taxes and transfer payments. This paper brings together all of these approaches to investigate the history of Black–White poverty differentials for families with children from 1980 to 2014. We break the history of the Black–White poverty differential into three “eras”: 1980 to 1992, when the racial differential was largely driven by the business cycle; 1992 to 2002, when the racial differential was reduced substantially; and 2002 to 2014, when the differential could have been strongly influenced by the Great Recession, but was not. For each era, we examine the extent to which the changes in the poverty differential were influenced by changes in tax and transfer payment policy and by changes in family demographic and labor market characteristics. We find that labor market changes and changes in tax credits and transfer payments have strongly influenced the differential, though racial differences in marital structure, family work effort, and heads’ educational attainment also continue to play a role.
Introduction
In his 1988 State of the Union Address, President Reagan declared, “My friends, some years ago the federal government declared war on poverty, and poverty won” (State of the Union Address, 1988). It is easy to understand why a critic today might hold the same opinion. If one looks at the graph of the official poverty measure (OPM) in Figure 1, it appears that the poverty rate among persons in families with children was essentially the same in 2014 as it was in 1982, and that the progress of the 1990s was mostly reversed in the subsequent years. This paper uses an improved measure of poverty, the anchored historical supplemental poverty measure (hereafter SPMa) introduced by Wimer et al. (2016), to draw a rather different set of conclusions. When compared to the OPM line, the SPMa poverty line in Figure 1 shows a more precipitous decline in poverty in families with children during the 1990s and a much smaller increase in poverty during the Great Recession period that followed.
A comparison of official and anchored supplemental poverty measures (SPM), 1980 to 2014.
Our primary research question is whether Black families were full participants in these favorable outcomes, particularly in comparison to White families—and why. Figure 2 shows the same comparison as Figure 1 for White and Black families separately, and Figure 3 plots the difference between Black and White poverty by both measures. We will analyze the results in Figure 3 by examining three time periods that we call “eras.” In Era 1 (1980–1992), the racial poverty differential had no obvious trend by the OPM, but declined somewhat by the SPMa, and was sensitive to the business cycle by both measures, with the racial poverty differential rising during recessions. Era 2 (1992–2002) was associated with a rapid decline in the racial poverty differential, whether measured by the OPM or SPMa. And in Era 3 (2002–2014), the OPM shows an increase in the Black–White poverty differentials, whereas the SPMa indicates that the differential continued to decline, even during the Great Recession. 1
Official versus anchored supplemental poverty measures (SPM) poverty for families with children by race, 1980 to 2014.
Black–White official versus anchored supplemental poverty measures (SPM) poverty rate differentials for SPM families with children, 1980 to 2014.
The stylized facts about a declining racial poverty differential using the SPMa could arise from various combinations of three basic factors: (a) convergence of Black and White parents’ educational attainment and success in labor markets; (b) convergence of family demographic traits like marriage rates or family size; and (c) changes in tax and transfer policies that served to narrow the racial differential in families’ ability to escape poverty. This paper investigates the importance of all three factors using the SPMa dataset and techniques of analysis that combine the approaches found in the most recent research. After reviewing this literature and describing our dataset, we will apply these techniques to analyze what happened in each era.
Background: The Macroeconomy and the Labor Market 1980 to 2014
Before beginning to examine each era separately, we provide a brief description of the U.S. labor market during the 1980 to 2014 time period to provide an economic context for our analysis.
Economists Holzer and Hlavac (2012) describe the U.S. labor markets of the past 30 years as a “Very Uneven Road of Progress,” exhibiting major structural changes and large swings in economic performance. The 1980s began with a severe economic recession engineered by the Federal Reserve Bank to reduce inflation. What followed was moderate growth in earnings driven primarily by growth in employment, as the Baby Boom Generation entered their prime earning years. The 1990s began with a mild recession followed by the “Roaring Nineties,” a period of robust economic growth, tight labor markets, and more substantial gains in wages and earnings. Over the period, strong consumer demand for goods and services pushed the unemployment rate to a 30-year low, whereas a change in emphasis from income support to work support via welfare reform and the expansion of the earned income tax credit (EITC) increased labor force participation among less educated single mothers. The 21st century began with a brief recession, but with a mixed recovery: higher productivity due to technological change, but a weaker employer demand for labor resulting in weak wage and earnings growth. The decade ended with the Great Recession (2007–2009), at the time the largest economic downturn since the Great Depression. Though production responded relatively quickly, it took labor markets nearly 6 years to fully recover the number of jobs lost.
Holzer and Hlavac (2012) go on to show that labor market performance was not only uneven during the period, but also that the gains and losses from these changes were unevenly experienced by different demographic groups. Over the past three decades, skilled-biased technical change has favored college-educated men and women who generally experienced larger gains in employment, wages, and earnings than did the less educated. Until the 21st century, regardless of education level, women experienced faster employment, and wage growth than did men. Combining these trends, younger, less educated, male workers fared particularly poorly over the 35-year period: Hourly wages for men with a high school diploma or less were stagnant for the entire period, and their earnings declined slightly.
Figure 4 documents unemployment rates for Blacks and Whites, age 16 and over during the period. Black unemployment rates are persistently higher than White unemployment rates, and Blacks are more susceptible to economic downturns. Cajner et al. (2017) examined Black–White disparities in key labor market outcomes over the period and find that Blacks have substantially higher and more cyclical unemployment rates than do Whites, not due to any observable characteristics, but instead driven by a comparatively higher risk of job loss. Couch and Fairlie (2010) find considerable evidence that Blacks are first fired as the business cycle weakens. The low labor force participation of Black men is not well-explained by observables, but is mostly driven by higher labor force exit rates from employment which have shown little improvement in the last 40 years. Among those who work, Blacks are more likely to work part-time than Whites, despite wanting more hours, and Black–White gaps in involuntary unemployment remain large even after controlling for differences in observable characteristics. Robust economic activity consistently reduces Black–White gaps in employment and wages, yet disparities remain substantial.
Unemployment rates for Blacks and Whites, age 16 and over, 1980 to 2015, annual averages.
Background: Previous Research About Black–White Poverty Differentials
For many years research on Black–White differentials focused on earnings of individual workers, not on family poverty per se. Farley (1984) was an exception, noting that while the Black–White poverty differential had decreased during the 1960s, it had ceased to decline in the 1970s, 2 but most of the book is devoted to the topics of educational attainment, employment, and earnings. Blackburn et al. (1993) and Darity and Myers (1998) expanded earlier research, emphasizing the role of improvements in the educational qualifications of the Black labor force. Couch and Daly (2002) went on to show that the 1990s narrowing of the Black–White poverty differential was primarily due to continuing improvements in the educational attainment and occupational standing of Black men. Sakamoto et al. (2018) summarized much of the subsequent research, again emphasizing improvements in educational attainment. They also provide evidence from longitudinal data that the results for low-skill Black men would be better yet if they were not affected by years with zero reported earnings, whether due to lack of success in the labor market or periods of incarceration.
It was not until the late 1980s that research turned more directly to the study of family poverty, especially child poverty. These studies relied upon the OPM, and the analysis proceeded by studying racial differences in observable traits of families or family members. Eggebeen and Lichter (1991) were the first to tie together racial differentials in both family structure (the prevalence of single-parent families) and parental work patterns (primarily maternal employment) with racial differentials in child poverty. They found that changes in parental work patterns had reduced racial differentials in child poverty, but changes in family structure had the opposite effect. Lichter et al. (2005) used a similar data design to investigate the decline of child poverty in the 1990s, showing that the decline was primarily a consequence of increased maternal employment rather than changes in family structure, whereas family structure differences continued to be a major factor in explaining the ongoing racial differential in child poverty.
Gundersen and Ziliak (2004) also investigated changes in OPM poverty during the 1980s and 1990s for White families and Black families separately, but focused on measuring the effect of macroeconomic trends using state-level panel data. They found that declining unemployment in a state had a strong effect on poverty among White families, but not Black families, whereas rising median wages in a state reduced poverty among Black families, but not White families. An additional component of this research was that it moved beyond the OPM by providing a separate set of results that looked at after-tax income and the impact of the EITC, finding that the effect of declining unemployment on poverty in Black families more closely resembled the effect on White families when poverty was measured using after-tax income. Bitler et al. (2017) conducted a similar analysis for the years 2000 to 2014, again using state-level panel data, but abandoning the OPM entirely. Comparing income after taxes and transfers to a historical SPM threshold, they find that the effect of 1 percentage point increase in unemployment is associated with a slightly larger increase in child poverty for Blacks (0.492 percentage points) than Whites (0.363 percentage points), but the difference is not statistically significant. It appears that the tax and transfer system has practically eliminated a racial differential effect of the unemployment rate on family poverty, though it is less clear about the relative importance of taxes versus transfer payments.
From this review, one might infer that there is a paucity of research that focuses directly on the history of the Black–White poverty differential in an analytical way. There are only three exceptions. Two of them emphasize Black–White differentials in observable traits of the family and its adult members (Gradín, 2012; Iceland, 2019), and the other one builds on what is to be learned from comparing poverty measures (Nolan et al., 2016).
The two papers that focus on observable traits both rely on the Blinder–Oaxaca decomposition technique that divides the racial poverty differential into two components: a component that is “explained” by racial differentials in the mean value of observable variables and an “unexplained” residual. 3 Gradín (2012) divides observable variables into two broad categories: “sociodemographic” and “education and labor activity.” Racial differences in the explanatory variables “explained” 84.6% of the racial poverty differential in 1993, 74.1% of the differential in 2001, and 76.6% of the differential in 2006 (the final year in Gradín’s dataset). The two broad categories were of roughly equal importance. In a detailed breakdown of 2006 results, the most important variables by far were (in order of the portion of the poverty differential explained) work effort of adults, number of children, marital status of head, age of head, and education of head.
Iceland (2019) uses a similar analytical design with two major differences: he relies on long-form Census data for 1959 and 1979 and American Community Survey data for 2015 rather than the Current Population Survey data used by Gradín, 4 and he does not use any variables measuring work effort. Racial differences in demographic variables “explained” 67% of the Black–White poverty differentials in 2015, the most important categories being (in order of proportion of the differential explained) marital status of head, age of head, and education of head. Family size differences, which had explained 9% of the differential in 1979, explained only 2% of the differential in 2015.
The timing of the research by Gradín and the dataset chosen by Iceland precluded them from using more contemporary measures of poverty. Nolan et al. (2016) are able to use the SPMa data 5 to investigate trends in child poverty. First of all, they directly compare the trends using the OPM to those using the SPMa just as we did in the introduction to this paper. Their comparison of the OPM and SPMa is startling: From 1970 to 2014 the Black–White differentials in child poverty measured by the OPM fell by 6.3 percentage points from 33.6% to 27.3%; the SPMa differential fell more than twice as much, by 14.5 percentage points from 33.1% to 18.6%. Their use of the SPMa allows them to use the SPMa thresholds to compare the poverty rates based on two different measures of household resources, income before taxes and transfers and income after transfers and taxes, much as Bitler et al. (2017) do, but using the new data in a different way. The comparison of poverty rates using the same threshold, but different income measures, provides a partial explanation of the declining differential: From 1970 to 2014 the Black–White differentials in child poverty based on pretax/transfer income decreased by only 5.9 percentage points from 37.3% to 31.4%; the SPMa differential fell by four times as much, 24.2 percentage points from 40.8% to 16.6%. These exercises show that using the SPMa reveals much more progress in reducing the Black–White poverty differentials than using the OPM or a measure that focuses exclusively on narrowing the gap in market income. However, neither exercise addresses the question of how much of the convergence in poverty rates was due to changes in the observable traits of the families.
Our research uses both the SPMa definition of family resources and the Blinder–Oaxaca decomposition procedure to investigate the relative role of family demographics, employment and labor market effects, and government taxes and transfers on the Black–White poverty differentials. We find that the relative importance of these factors changes over time and varies significantly with macroeconomic conditions and the policy environment.
The next two sections of this paper describe the data and our methods of analysis. We then apply our methods to each era separately to interpret the stylized facts of each time period. The research concludes with a summary of the principal findings, including potential implications for public policy.
The Data
To conduct our analysis, we merge data from 1980 to 2016 March supplement to the Current Population Survey (ASEC) with the same historical SPMa poverty data used in Wimer et al. (2016). Table 1 describes the key differences that explain why the SPMa is superior to the OPM for looking at trends in family poverty. Two of the differences are technical improvements based on the recommendations in Citro and Michael (1995): an improved version of the consumer price index is used to adjust for inflation, and an improved version of the “equivalence scales” is used to adjust the poverty thresholds for family size. Three other changes are more substantive. First, the SPMa incorporates nonfamily members in the measurement unit (hereafter called the “SPM family”), which now includes unmarried partners and their relatives as well as coresident children and foster children, adapting to changing trends in family structure (Provencher, 2011). Second, changes in the thresholds are driven by the cost of food, clothing, housing, and utilities, not just food, and the required expenditures to avoid poverty are based on the actual expenditures on these by families. 6 The third difference is the most important: Tax credits and noncash transfer payments are added to income, whereas taxes paid, work and child care expenses, and out-of-pocket medical costs (medical out-of-pocket expenditures [MOOP]) are subtracted, giving a more exact measure of the resources that are actually available to meet family needs (Blank, 2008). To assure comparability over time, the SPMa is anchored at 2012 levels of expenditures on food, clothing, housing, and utilities and adjusted for inflation.
A Comparison of the Official Poverty Measure (OPM) and an Anchored Supplemental Poverty Measure (SPMa).
The measures use slightly different versions of the consumer price index.
Necessities include food, clothing, housing, and utilities.
Our dataset is restricted to SPM families containing at least one child under age 18. In order to better analyze subsamples, a “year” in our data is a 3-year moving average centered on a particular calendar year. For example, variable values for 1991 are based on data for calendar years 1990, 1991, and 1992 (ASEC survey years 1991, 1992, and 1993). The principal advantage is that the resulting datasets are large—always more than 47,000 White families and 6,500 Black families—assuring that almost any differential will be statistically significant. 7
In defining the variables used in our analysis we have generally made decisions consistent with those made by Wimer et al. (2016) in constructing the historical SPMa. The most important example is the identification of cohabiting partners of unmarried family heads. Cohabitating partners were first identified in the ASEC in 1995; in order to find cohabitating couples for previous years, we applied the partners of opposite sex sharing living quarters (POSSLQ) methodology described in Fox et al. (2015). Thus, we consider children to be in married couple, single parent, or cohabiting couple families based on the status of the SPM family head.
We limit our analysis to two racial groups based on the racial identity of the SPMa family head: “White alone,” or “Black alone.” “Families” refers to SPMa families, and the phrase “all families” means Black and White families combined. As such, the families we study are not exhaustive of all poor families with children, ignoring, for example, Asian families and their children. 8
Table 2 contains the values of the observable characteristics of families and family heads at all of the era endpoints. For purposes of reporting, we group the explanatory variables into six categories. Three of the variables represent family characteristics: number of persons (separately for adults and children), usual weekly hours worked per adult age 18 to 64, and census region (including metropolitan status). The other three represent characteristics of the head: age (and its square), marital status (with cohabitation as a separate status), and educational attainment in four exhaustive categories (less than high school, high school only, some college, and college graduate).
Demographic Characteristics of SPM Families with Children, All and by Race, Select Years 1980 to 2014.
Note. SPM = supplemental poverty measure.
Table 2 shows that there has been a trend away from marriage among families with children. The declining percentage of family heads who are married has meant an increase in the number of nonpartnered single heads, but also a rapid increase in the percentage of heads who have cohabiting partners. Unmarried heads have been much more common among Black families than White families throughout the period we are examining, though the trend towards fewer married heads has been similar in both racial groups. The trend towards cohabiting heads has been somewhat greater for White families than Black, to the point that the percentage of families in that category is now almost the same. Although overall family size is very similar for both groups, White families have slightly fewer children and slightly more adult members than do Black families. There has been very little trend in the usual hours worked by a family, and a racial differential of about 5 h per adult has remained roughly constant. 9 The trend towards greater educational attainment has been characteristic of both groups; in particular, the share of family heads who lack a high school education has declined rapidly among Blacks and Whites alike. College attendance and college graduation have increased more rapidly among Black family heads than White, but White heads are still almost twice as likely as Black heads to have achieved a college degree. The last trend that emerges from Table 2 is the rapid decline in the rural (nonmetropolitan) population, which has been reduced by half since 1980 among both Black and White families. The question is how much of the racial differentials in poverty are explained by differentials in factors such as marital status, hours of work, and educational achievement? We will find some answers in the analysis that follows, but first, we must explain our methods.
Methods
A major advantage of employing the SPMa is that each SPMa family’s private income, taxes (including tax credits), government transfers (both in cash and in kind), and relevant household expenses are separately calculated. From there we can calculate a series of poverty measures that use a common unit of analysis (the SPMa family) and poverty threshold (SPMa) while focusing on different resource measures, as shown in Table 3. 10 For each era, we first analyze the alternative poverty measures described in Table 3 for Black and White families combined to identify key changes or trends in the labor market and policy environments during the particular era. We then complete the same analysis of alternative poverty measures for Black and White families separately, revealing the extent to which the key changes and trends in the labor market and policy environments affected Black–White poverty differentials. Finally, we use a Blinder–Oaxaca decomposition procedure to determine the extent to which Black–White poverty differentials can be further explained by differences in observable traits of families and family heads at both the beginning and the end of each era. The Blinder–Oaxaca decomposition is a statistical procedure that enables us to explain a portion of the SPMa poverty differential between Blacks and Whites by decomposing that difference into the part that is “explained,” by differences in the mean values of specific observable characteristics of Blacks and Whites (demographics, educational attainment, and work effort) and an “unexplained” residual (see the Supplemental Appendix for technical details). By using this sequence of techniques, we are able to integrate the approaches found in the previous literature to provide a more comprehensive understanding of how and why the Black–White poverty differential has evolved.
Alternative Measures of Poverty, 1980 to 2014.
Note. SPMa = anchored supplemental poverty measure; AFDC = Aid to Families with Dependent Children; TANF = Temporary Aid for Needy Families; WIC = Special Supplemental Nutrition Program for Women, Infants, and Children; SPM = supplemental poverty measure; MOOP = medical out-of-pocket expenditures.
Results and Discussion for Era 1: 1980 to 1992
Figure 5 shows the various poverty measures for Era 1 for Black and White families combined. The most obvious characteristic is that all of the poverty measures show considerable sensitivity to the business cycle, with poverty peaks representing the recession periods of the early 1980s and early 1990s. As discussed above, the sensitivity of poverty to the business cycle has been a topic of research since Gundersen and Ziliak (2004), but the topic of racial differentials has not usually played a role in that conversation.
Alternative definitions of poverty for Black and White families with children, Era 1: 1980 to 1992.
In Table 4, we present some basic results using our data. The top panel shows the results of a very basic regression to measure the percentage points of change in poverty per 1% change in the annual state unemployment rate with only the year and region as control variables. The private income poverty columns show the poverty effect in the absence of the safety net tax and transfer programs, and the SPMa poverty columns show the effect with safety net benefits included. The results for Era 1 show exactly what Figure 5 leads us to expect, that the impact on SPMa poverty is smaller than the impact on private income poverty. For example, the national unemployment rate rose by 2.1 percentage points, from 5.3% to 7.4%, between 1988 and 1992, so we would expect the private income poverty rate to rise by 3.1 percentage points, but the SPMa poverty rate to rise by only 2.6 percentage points, and that is just what Figure 5 shows. As shown in Table 2, however, there were demographic changes during Era 1 that were also affecting these poverty rates, so the bottom panel of Table 4 incorporates controls to isolate the effect of the unemployment rate, and indeed the coefficients on the unemployment rate are now smaller. Using the same example of the 2.1 percentage point increase in the unemployment rate, the poverty impact attributable to the unemployment rate would be an increase of 2.2 percentage points in the private income poverty rate and 1.8 percentage points in the SPMa poverty rate. Fully understanding these differentials requires that we explore the safety net changes in Era 1 and their effect on the SPMa poverty.
Change in Poverty Rate per 1% Change in Annual State Unemployment Rate.
All coefficients are significantly >0 at significance level .001.
All coefficients for Blacks differ from those for Whites at significance level .001.
Region controls include census region and metropolitan status.
Demographic controls include number of adults and children, head’s marital/cohabitation status, and head’s age and education level.
Changes in safety net policy are reflected in a distinctive characteristic of Era 1 in Figure 5: After-tax and post-transfer poverty diverges from private income poverty, especially during the recession of the early 1990s. The reasons are shown in the top panel of Table 5, which contains recipiency rates and real mean benefits received by all private income poor families.
Era 1: Recipiency Rates and Real Mean Benefits Received, Private Income Poor Families with Children Under 18 (2002 dollars).
Note. AFDC = Aid to Families with Dependent Children; SPM = supplemental poverty measure; SNAP = Supplemental Nutrition Assistance Program.
Table 5 is the first of a set of tables, one for each era, that shows the recipiency rate for important income sources and the average amount received by those recipients. 11 The top panel of Table 5 shows that EITC recipiency was somewhat sensitive to the business cycle, and that the real value of EITC benefits for recipients increased substantially across Era 1 due to major policy changes. The EITC was initially introduced in 1975, but the benefits were not indexed, and consequently declined in real terms. The Tax Reform Act of 1986 restored the purchasing power to a little more than the 1975 level and indexed the key parameters. In addition, the phase-in and phase-out rates were made more generous in 1986, 1988, and 1990, and more generous schedules for families with two or more children were introduced in 1990 (Nichols & Rothstein, 2016).
Unlike the EITC, the Food Stamp Program had no major policy changes in Era 1 (Hoynes & Schanzenbach, 2016; Ziliak, 2016b). By the end of Era 1, however, food stamps had become quite sensitive to economic conditions, as the sharp increase in 1992 shows (see also Figlio et al., 2000; Ziliak et al., 2003). Table 5 also shows the major increases in housing benefit recipiency. Growth in both tenant and project-based Section 8 housing programs and the introduction of the Low-Income Housing Tax Credit as part of the 1986 Tax Reform led to real growth (Collinson et al., 2016)
The top panel of Table 6 shows two sets of expenditures that are subtracted from a family's after-tax and transfer payments income when calculating SPMa poverty: work expenses and child care costs, and MOOP. 12 The percentage of families with expenses for work and child care declined slightly during Era 1, and the average amount spent by those who had such expenses went up slightly; the trends in MOOP were similar, but the decline in number paying went down by a greater amount and the amount spent rose more. It is clear from Figure 5 that these expenses are far from negligible, as the SPMa poverty rate is almost 5 percentage points higher than the after-tax and post-transfer poverty rates throughout Era 1.
Era 1: Payment Rates and Real Mean Expenses, Private Income Poor Families with Children (2002 dollars).
Note. SPM = supplemental poverty measure; MOOP = medical out-of-pocket expenditures.
Era 1 began with a 27.3 percentage point differential in Black–White SPMa poverty and ended at almost the same value, a 28.2 point differential. These differentials were not constant across the time period, however. Figure 6a and 6b shows the same poverty measures as Figure 5, but for White families and Black families separately. The profiles are similar in shape because they are largely driven by the business cycles of Era 1. As the results in Table 4 lead us to expect, the changes in poverty over the course of a recession are larger for Black families than White families. In other ways, Figure 6a and 6b is rather different, however. For the White families, SPMa poverty is higher than private income poverty and tracks after-tax poverty closely, whereas for Black families SPMa poverty is lower than after-tax poverty and in the latter years lower than private income poverty. These differences are further explored in the bottom two panels of Table 5. White families were more likely to have a private income, with corresponding differentials in work-driven tax credits (EITC) and social insurance receipts (unemployment insurance and workers’ compensation), whereas Black families were more likely to receive means-tested transfers. A notable trend during Era 1 is that there was a convergence in participation in the Aid to Families With Dependent Children (AFDC) and Food Stamp Programs, as participation by White private income poor families rose substantially, whereas participation by Black private income poor families did not. Participation in housing and energy assistance increased for both Black and White families.
(a) Alternative definitions of poverty for White families with children, Era 1: 1980 to 1992. (b) Alternative definitions of poverty for Black families with children, Era 1: 1980 to 1992.
The bottom two panels of Table 6 provide information on White and Black family expenditures on child care and work expenses and on out-of-pocket medical expenses respectively. White families were more likely than Black families to have expenses for work and child care, and after 1980 Whites also had higher expenditures; MOOP expenditures, on the other hand, were very similar in both racial groups.
Table 7 brings together the analytical tools that were employed by Nolan et al. (2016), Gradín (2012), and Iceland (2019) to understand the relative roles of taxes and transfers and of the demography and labor market behavior of families in determining the Black–White differentials in SPMa poverty. The top panel of Table 7 shows how the differential in SPMa poverty is derived from the differential in private income poverty, whereas the bottom panel uses the Blinder–Oaxaca decomposition technique to calculate how much of the SPMa poverty differential is “explained” by each of several observable traits of families and family heads.
Era 1: Derivation and Blinder–Oaxaca Decomposition of Racial Differential in SPMa Poverty 1980 and 1992.
Note. SPMa = anchored supplemental poverty measure; SPM = supplemental poverty measure; MOOP = medical out-of-pocket expenditures.
The top panel of Table 7 shows that the tax and transfer payment changes during Era 1 reduced private income poverty by 7.6 percentage points more for Black families than White families in 1980 and by 7.9 percentage points more in 1992. These large differentials were partially offset by the effects of expenditures on work, child care, and out-of-pocket medical expenses, 13 but still sufficient to reduce the SPMa poverty differential by 4.9 percentage points in 1980 and 4.7 percentage points in 1992.
The bottom panel of Table 7 turns to the question raised by Gradín (2012) and Iceland (2019): To what extent can racial differentials in observable traits of the families and family heads explain the racial differential in SPMa poverty? And did the demographic and work effort changes during Era 1 alter the relative importance of any of those traits? 14 The details of the Blinder–Oaxaca decomposition technique are described in the Supplemental Appendix, but the basic idea is to measure how much the SPMa poverty differential would have been reduced if the White and Black populations had identical observable traits, referred to as the “explained” share of the actual differential. A positive value shows the number of percentage points that the differential would be reduced if the mean values of that trait were identical, and a negative value indicates that the differential would have been slightly increased if Blacks and Whites had identical traits. As shown in the bottom panel of Table 7, six variables “explain” 64.1% of the racial differential in 1980 and 66.3% of the differential in 1992. 15 In 1980, the most important factor was the marital status of the head, the interpretation being that if the shares of Black and White family heads who were married or cohabiting or unpartnered were identical, the SPMa poverty differential would be reduced by 7.9 percentage points, nearly 30% of the Black–White poverty differential and more than the shares of usual weekly hours worked per adult and educational attainment of the head combined. The importance of marital status had declined somewhat by 1992, whereas the importance of usual hours worked was substantially greater than it had been in 1980, as one might expect in a recession year. Educational attainment had also increased in importance, similar to the results of improved educational attainment that were emphasized by Darity and Myers (1998) in their analysis of racial differentials in earnings.
Summing up, the analysis of Era 1 has reached four conclusions that should be kept in mind when proceeding to the next two eras. First, private income poverty is more sensitive to the business cycle than SPMa poverty, especially for Black families. Second, there were trends towards the expansion of the generosity of the EITC and participation in the food stamp and housing assistance programs, but racial differences in participation did not generate a large change in the effect of taxes and transfers on poverty. Third, the safety net for White families was skewed towards work-related components, whereas the safety net for Black families was skewed towards means-tested transfer payments, though the racial differential in recipiency became smaller for both sets of programs. And finally, SPMa poverty for both White and Black families ended the era at almost the same level as it began. The lack of change in the Black–White poverty differentials was largely a consequence of the fact that the private income Poverty differential actually increased slightly. A Blinder–Oaxaca decomposition based on observable traits shows that the most important explanatory variables for the differential in SPMa poverty were, in order of importance, marital status of the head, hours of work per adult, and educational attainment of the head, with educational attainment becoming more important by the end of the era and marital status becoming less important.
Results and Discussion for Era 2: 1992 to 2002
Era 2 differed from Era 1. During Era 2 the SPMa poverty rate fell by 7.5 percentage points from 19.8% to 12.3%. Why did SPMa poverty fall so fast? Figure 7 provides useful clues.
Alternative definitions of poverty for Black and White families with children, Era 2: 1992 to 2002.
Figure 7 shows that during Era 2, private income poverty declined significantly, the interpretation being that Era 2 was (until the very end) a period of declining unemployment rates and rising private income, with declining private income poverty as a result. Figure 7 also shows that the SPMa poverty generally followed a path parallel to after-tax private income poverty. Prior to 1996, after-tax private income poverty exceeded private income poverty, but after 1996 tax credits caused the tax system to have the overall effect of reducing poverty, as it does to this day.
The obvious explanation for the anti-poverty impact of the tax system is the continued importance of the EITC, which played an important role in the rapidly falling rate of SPMa poverty characteristic of the early 1990s. In addition to the changes that occurred in Era 1, the EITC credits for families with two or more children, introduced in 1990, were roughly doubled in 1993 (Nichols & Rothstein, 2016), almost at the beginning of Era 2. Looking at the top panel of Table 8 we can see that EITC recipiency increased by more than 9 percentage points between 1992 and 2000 (though it declined somewhat during the subsequent recession), and more important yet, the real amount received by those receiving it doubled. Given the complex design of the EITC and its equally complex set of incentives (increasing the incentives to work for single parents, whereas decreasing the incentives for second earners in married-couple families with children), it is not clear how much of this dramatic change was due to policy changes, how much to changes in labor market opportunity/behavior, and how much to the interaction between those factors, but the magnitudes certainly suggest that the early 1990s expansions of EITC played a role (Ellwood, 2001). There is a consensus in the literature that EITC expansion led to sizable increases in single mothers’ employment rates, especially for less educated mothers with more than one child (Nichols & Rothstein, 2016, p. 188). Ventry (2001, pp. 37–38) summarizes the results of the EITC changes in this way: “By 1996, the EITC’s responsibilities had changed in one important respect: it was all that kept millions of individuals out of poverty.”
Era 2: Recipiency Rates and Real Mean Benefits Received, Private Income Poor Families with Children Under 18 (2002 dollars).
Note. AFDC = Aid to Families with Dependent Children; SPM = supplemental poverty measure; SNAP = Supplemental Nutrition Assistance Program; TANF = Temporary Aid for Needy Families.
Table 8 also shows results for the most important means-tested programs during Era 2. The recipiency rate for AFDC/Temporary Aid for Needy Families (TANF) dropped continuously and substantially during the 1992 to 2002 period, and the mean real transfer to recipient families also fell. Of course, 1996 was the year in which the Personal Responsibility and Work Opportunity Act (PRWORA) welfare reform was enacted, preceded in several states by program innovations implemented through waivers. This welfare reform converted the AFDC entitlement into a block grant system known as Temporary Aid for Needy Families (TANF). It also implemented both a work requirement and a lifetime benefit limit, thus emphasizing work support over income support. Employment among single mothers increased substantially, though there is an unresolved debate about how much of this is to be attributed to a tight labor market with increasing real wages, how much to the increased generosity of the EITC, and how much to the implementation of the TANF restrictions. 16 During Era 2, food stamps and housing and energy assistance recipiency also dropped substantially though it remains unclear whether these changes were simply a consequence of higher earnings by potential recipients or a consequence of the tightening of regulations associated with welfare reform.
We also see in the top panel of Table 9 that increased employment was associated with a corresponding increase in work and child care expenses, and MOOP expenditures increased also. Both the decline in transfer payments and the increase in expenditures partially offset the poverty reduction generated by higher earnings.
Era 2: Payment Rates and Real Mean Expenses, Private Income Poor Families with Children Under 18 (2002 dollars).
Note. SPM = supplemental poverty measure; MOOP = medical out-of-pocket expenditures.
In addition to the dramatic drop in SPMa poverty, the second characteristic of Era 2 was that the SPMa poverty rate for Blacks fell much more (14.7 percentage points, from 42.5% to 27.8%) than for Whites (6.1 percentage points, from 15% to 8.9%). The SPMa poverty rate for Blacks was much larger than that of Whites, to begin with and had much more room to fall, but this was still a notable drop in the racial SPMa poverty differential, from 27.5 to 18.9 percentage points.
Looking at the lower two panels of Table 8, we see that the percentage of the private income poor who had nonzero private income rose for both Blacks and White families, but rose much faster for Black families, and that there was a correspondingly rapid increase in EITC participation. It is not surprising, therefore, that after-tax private income poverty fell for both racial groups, but more for Black families than White. Offsetting that reduction in the Black–White differentials was the rapid increase in work and child care expenses for Black families (as shown in the bottom panel of Table 9), presumably as a by-product of increased employment.
The most straightforward inference from the after-tax and post-transfer results in Figure 8a and 8b is that the share of families removed from poverty by adding transfer payments to the income package remained much larger for Black families than for White families. Comparing the two bottom panels in Table 8, we can see how these racial differences evolved: Participation in means-tested transfers remained higher among Black families than White, though the recipiency rates of AFDC/TANF converged substantially.
(a) Alternative definitions of poverty for White families with children, Era 2: 1992 to 2002. (b) Alternative definitions of poverty for Black families with children, Era 2: 1992 to 2002.
In the years leading up to the 1996 welfare reform, AFDC had long been stereotyped as differentially beneficial to the Black population, and we saw in Table 5 that AFDC was received by >50% of the Black private income poor families throughout Era 1, compared to <40% of the White private income poor families. The rapid drop in recipiency shown in Table 8 is presumably a result of the welfare reform, as it eliminated a major source of income for many families, particularly Black families. Era 2 also saw a reduction in food stamps and housing and energy assistance recipiency, a reduction in participation of about 15 percentage points in both racial groups.
We draw two overall conclusions from the analysis of racial poverty differentials in Era 2. First, improving labor markets, reinforced by the EITC, were a primary driver of the racial convergence in SPMa poverty levels. 17 This interpretation is also reflected in Table 4, where the sensitivity of poverty rates to the declining unemployment rate was twice as large for Blacks as for Whites in Era 2 even after accounting for changes in demographics. Second, despite the rapid decline of AFDC/TANF, transfer payments as a whole (particularly food stamps) remained important in relieving poverty, especially in Black families.
Table 10 shows the extent to which the relative roles of taxes and transfers and the observable traits of families and their heads changed during this period of a rapidly declining racial differential in SPMa poverty. The top panel in Table 10 shows that differences in the effect of taxes and transfers continued to play an important role in explaining SPMa poverty differentials, but the size of that effect scarcely changed between 1992 and 2002. The bottom panel in Table 10 shows that usual hours per adult was far less important in 2002 than it had been in 1992, as one would expect during a period of low unemployment.
Era 2: Derivation and Blinder–Oaxaca Decomposition of Racial Differential in SPMa Poverty 1992 and 2002.
Note. SPMa = supplemental poverty measure anchored; SPM = supplemental poverty measure; MOOP = medical out-of-pocket expenditures.
We complete our exploration of the decline of the racial differential in Era 2 by asking whether that decline is partly attributable to changes in the observable traits of families and their heads. The demographic data shown for Era 2 in Table 2 showed three trends. First, the decline in the percentage of households with married heads observed in Era 1 continued in Era 2, with a corresponding increase in cohabiting households. Second, usual weekly hours per working-age adult increased somewhat (by 1.3 h) during this period of falling unemployment. Third, the educational attainment of the heads increased substantially, with the percentage of heads with at least some college rising from barely 50% to almost 60%. Both the work effort and educational attainment trends were stronger for Black households than Whites. The Blinder–Oaxaca decomposition analysis in the bottom panel of Table 10 shows that the role of these variables in “explaining” the racial gap in SPMa poverty was smaller in 2002 than in 1992, exactly what one would expect from the racial convergence of work effort and educational attainment across the period. It raises the question of whether it is possible to provide a similar decomposition of the change in the Black–White poverty differentials.
Since the effect of taxes and transfers changed so little between 1992 and 2002, the decline in the SPMa poverty differential must be “explained” to some extent by changes in the observable characteristics used for the Blinder–Oaxaca decompositions. Table 11 shows the results of a novel decomposition 18 of the change in the SPMa differential that shows how much of the smaller differential can be “explained” by the convergence of the key factors. The most important convergence was in weekly hours worked per prime-age adult, followed by educational attainment, and head’s marital status. Racial convergence in these three factors combined accounts for 61.6% of the racial convergence in SPMa poverty.
Blinder–Oaxaca Decomposition of Change in Racial Differential of Anchored SMP Poverty, 1992 and 2002.
Note. SPMa = supplemental poverty measure anchored; SPM = supplemental poverty measure; MOOP = medical out-of-pocket expenditures.
Summing up our results for Era 2, our two big questions were: Why did SPMa poverty fall so rapidly? And why did the Black–White differentials in SPMa poverty also fall by such a large amount? We conclude that Era 2 was a period of rapid reduction in SPMa poverty due to interactions among at least four major factors. First, there was a rapid decline in unemployment, falling to a level that was not achieved again until 2018. Second, it was a period of rising real wages, spurred in part by improving the educational attainment of family heads, White and Black alike. Third, welfare reform strongly discouraged long-term reliance on AFDC/TANF, the principal cash welfare program for low-income families with children, and strongly encouraged poor families to increase labor earnings, which they did. Finally, the structure of taxes and transfer payments relied increasingly on the EITC as an anti-poverty program, so that after-tax poverty rates consistently fell even faster than private income poverty rates. Measured by percentage points of poverty reduction, all of these factors benefited Black families differentially. The fact that the rapid reduction in poverty in Era 2 was especially large in the Black population was already known from the OPM data, receiving brief mention in surveys of poverty history such as Haveman et al. (2014), but the analysis of SPMa poverty sheds a different light on this literature. In particular, our results suggest that improved educational attainment and the expansion of the EITC served to reinforce the decline in poverty driven by increased parental employment and earnings, but also that a robust transfer payment system retained a powerful role in reducing poverty, especially for Black families.
Results and Discussion for Era 3: 2002 to 2014
The SPMa poverty measure tells a rather surprising tale about Era 3. When we first looked at the history of SPMa poverty in Figure 1, we found that the rapid decline in SPMa poverty in Era 2 came to a halt in Era 3. But we also saw that the impact of the Great Recession was much smaller than the impact of the Era 1 recessions, so in spite of the Great Recession, the SPMa poverty rate ended up about where it began. Figure 3 went on to show that the Black–White differentials continued their downward trend despite the Great Recession. 19
Broadly speaking, then, there are three puzzles we wish to address. First, why did the rapid progress against poverty that characterized the 1990s practically cease, even prior to the Great Recession? Second, why was the impact of the Great Recession on SPMa poverty so small relative to the magnitude of the unemployment shock? And third, why did the SPMa poverty measure continue to show a degree of racial convergence?
Figure 9 is helpful in understanding why the rapid decline in SPMa poverty ceased in Era 3. It shows that private income, before and after inclusion of the EITC and tax effects, seems to have stopped being a factor in SPMa reduction even before the Great Recession, so the end of the falling SPMa poverty rate that characterized Era 2 was presumably a labor market phenomenon. This result is partly due to the fact that the unemployment rate fell only by 1.6 percentage points between 2002 and 2006, never reaching the levels of 1999 or 2000. It also reflects the fact that real wage increases in Era 3 were much smaller than had been characteristic of Era 2. As the data in Gould (2018, Supplemental Appendix Figure C) show, real wages at the 10th percentile were essentially the same in 2014 as they were in 2002.
Alternative definitions of poverty for Black and White families with children, Era 3: 2002 to 2014.
The second question is why SPMa poverty was so unresponsive to the sharp increases in unemployment associated with the Great Recession. We can address this question rather quickly because the phenomenon is well known and has been carefully investigated within the last few years by Bitler and Hoynes (2016), by Bitler et al. (2017), and also in a chapter on the effect of transfer payments on child poverty by Pilkauskas and Garfinkel (2016).
Our contribution can be inferred from the poverty rates in Figure 9: from 2006 to 2011, private income poverty rose by 3.9 percentage points, from 14.1% to 18.0%; after-tax private income poverty rose by 2.5 percentage points, from 12.8% to 15.3%; but after-tax and post-transfer income poverty rose by only 0.5 percentage points, from 8.1% to 8.6%. We seem to be looking at the extraordinary success of the safety net, as the recipiency data shown in the top panel of Table 12. Between 2006 and 2011, the share of the private income poor families receiving unemployment compensation or worker’s compensation nearly doubled. The portion receiving Supplemental Nutrition Assistance Program (SNAP) (food stamps) also went up by almost 10 percentage points, and the mean benefit increased due to an increase in the maximum allowable amount. The unemployment insurance system was temporarily expanded to extend the number of weeks that benefits could be received and to shift more of the expense onto the Federal budget, and family incomes were also increased by other program components. 20 Another potentially important policy change is reflected in the top panel of Table 13 showing the decline of MOOP expenditures after the passage of the Affordable Care Act in 2010. Together these changes seem to have been remarkably effective despite the fact that TANF recipiency actually fell during the Great Recession.
Era 3: Recipiency Rates and Real Mean Benefits Received, Private Income Poor Families with Children Under 18 (2002 dollars).
Note. SPM = supplemental poverty measure; SNAP = Supplemental Nutrition Assistance Program. TANF = Temporary Aid for Needy Families; SPM = supplemental poverty measure.
Era 3: Payment Rates and Real Mean Expenses, Private Income Poor Families with Children (2002 dollars).
Note. SPM = supplemental poverty measure; MOOP = medical out-of-pocket expenditures.
The third puzzle is why the Black–White poverty differentials continued its decline in a period that was dominated by a recession. Figure 10a and 10b shows that White SPMa poverty actually increased slightly across the era, from 8.9% to 9.5%, whereas Black SPMa poverty recovered well enough after the Great Recession that it fell from 27.8% to 25.1%. Thus, the SPMa racial differential fell slightly but steadily until 2010, and ended about 3 percentage points lower than when Era 3 began.
(a) Alternative definitions of poverty for White families with children, Era 3: 2002 to 2014. (b) Alternative definitions of poverty for Black families with children, Era 3: 2002 to 2014.
Figure 10a and 10b also shows that the safety net functioned particularly well for Black families during the period from 2006 to 2011. For Whites, private income poverty rose by 3.4 percentage points, but SPMa poverty rose by only 1.2 percentage points. For Blacks, private income poverty rose by 5.5 percentage points, but SPMa poverty actually fell by 0.6 percentage points. This suggests that we will find the explanation in recipiency of transfers and in the payments for work, child care, and medical expenses. The bottom panels of Table 12 show that there was a convergence in the percentage of families who were private income poor who received EITC and unemployment compensation, whereas Black families continued to be more likely than White families to receive SNAP benefits and housing assistance. A comparison of the bottom two panels of Table 13 also shows the Black families had a much larger decrease in MOOP expenditures after 2010.
The final evidence concerning the decline of the racial differential in SPMa poverty during Era 3 comes from the results shown in Table 14 showing the relative importance of taxes and transfers and of observable traits in explaining the racial differential in SPMa poverty. The top panel summarizes the conclusions that we have already reached: that Era 3 actually showed an increase in the private income poverty differential by 1.3 percentage points, but it was more than offset by the tax and transfer system, which reduced the SPMa poverty rate differential by 11.3 percentage points in 2014 compared to 8.1 percentage points in 2002. The lower panel of Table 14 shows the role of the observable characteristics that also “explain” a portion of the differential. Table 2 shows that Era 3 was a period of improving educational attainment for both Black and White family heads, particularly the share with a college degree, and the decomposition suggests that the gap in the educational attainment of the head has become less important in explaining the differential in SPMa poverty. The usual weekly hours per adult became slightly more important in explaining the differential, as one might expect in a period with a recession, but the largest observable trait that continues to explain the Black–White poverty differential is the racial differential in marital status.
Era 3: Derivation and Blinder–Oaxaca Decomposition of Racial Differential in Anchored SPM Poverty, 2002 and 2014.
Note. SPMa = supplemental poverty measure anchored; SPM = supplemental poverty measure; MOOP = medical out-of-pocket expenditures.
Summing up, Era 3 seems to carry three lessons. First, private income poverty, before and after taxes, is still driven by the labor market. When the labor market stagnates, so does SPMa poverty for Black and White families alike, because the tax and transfer systems are not designed to address stagnant labor markets. Second, the transfer payment safety net proved to be remarkably effective at mitigating the increase in poverty that one would have expected from the Great Recession because the transfer payment system was effectively adapted to that purpose. And third, a reduction in the Black–White poverty differential is partly a matter of improving trends towards higher educational attainment and increased job opportunities for Black parents, but would also benefit from changes in both the labor market and the benefits structure that could be more effectively crafted to the needs of unmarried family heads.
Conclusion
In 1980, at the beginning of our sample period, the SPMa poverty rate was 15.2% for White families with children and 42.5% for Black families with children, a differential of 27.3 percentage points. By 1983, a recession year, it had risen to a 30.4-point differential, but has never been above 30 points again. By 2002 the SPMa rates were 8.9% for White families and 27.8% for Black families; the differential had been cut to less than 20 percentage points. By 2010 the Great Recession had slightly increased SPMa poverty among White families to 9.3%, but SPMa poverty among Black families reached a new low at 24.0%, and the differential had fallen to less than 15 percentage points. The racial differential stands at 15.6 percentage points at the end of our sample period in 2014.
Looking at the sample period from beginning to end, then, the poverty differential fell by 11.7 percentage points. Our analysis has provided insight into how and why that differential has been reduced. As was obvious from Figure 3, most of this reduction occurred in Era 2, when the differential fell by 8.6 percentage points, driven by a 9.8-point reduction in the private income poverty differential. Declining unemployment, increased educational attainment, and a policy environment that rewarded work had combined to create an environment in the late 1990s that led to increased work effort and private income. By 2002 the racial differential in recipiency rates of private income and EITC among private income poor families was reduced by more than half compared to 1980, and by 2014 those differences in recipiency had been all but eliminated. Decreased unemployment also led to longer work histories that enhanced the access of Black workers to unemployment insurance during the subsequent recessions. Meanwhile, the continuing increase in the share of White families with an unmarried family head played a role in their increased reliance on means-tested transfer payments, particularly SNAP. 21 These changes together have led to a decline in the racial differential in SPMa poverty that continued right through the Great Recession, the first recession period that did not increase the differential.
In 2014 taxes and transfers reduced the 24-point racial differential in private income poverty by over 11 percentage points, but there was still a remaining gap of 15.6 percentage points in SPMa poverty. A naïve reading of the Blinder–Oaxaca decomposition for 2014 found in Table 14 suggests that the large racial gap in the percentage of heads who are married (35.8% for Black heads, 71.9% for Whites) “explains” 4.4 percentage points of the gap, and the difference in working hours per adult (24.9 h per week for Black families, 30.1 h per week for Whites) “explains” another 2.9 percentage points. The direction of causality is far from clear, however. Noting that Era 1 (1980–1992) and Era 3 (2002–2014) are about one generation apart, racial differentials in family structure and work effort per adult, as well as the remaining racial gap in educational attainment, are quite plausibly as much a consequence of poverty in the previous generation as a cause of poverty in the current generation. Corroborating evidence for this view is found in the results in Smeeding (2016a, Table 1) showing the likelihood that a child who was in a bottom quintile family in 1998 would be an adult in a bottom quintile family in 2010. This outcome was twice as likely for a Black family (51%) as a White family (23%), and similar disparities were associated with having a mother who was never married or a parent who was a high school dropout. These disparities are generally associated with family instability, low income and wealth, inadequate schooling opportunity (including preschool), higher probabilities of incarceration, and life in a disadvantaged neighborhood 22 ; and every one of those factors reduces the likelihood that the children will escape poverty when they become adults. Reducing the poverty differential in future generations is partly a matter of policies that narrow the differential for the current generation.
Our analysis contains implications for contemporary policy debates because it identifies policies that have been effective. First, the reduction of poverty through higher earnings requires changes in the labor market that provide more and better job opportunities for parents, who often need more flexible hours and broader benefits, and the wages for those jobs need to be high enough that when added to tax credits they are sufficient to escape poverty. Second, it is clear that the anti-poverty benefits of high employment rates can be enhanced by a liberalization of the EITC, and that the impact of child care expenses on the SPMa poverty could be mitigated by making the child care tax credit refundable. Neither of those changes was part of the 2017 tax legislation. Third, while increased labor force participation and higher wages are important, they are not sufficient to eliminate poverty. Even families with substantial earnings are often in need of means-tested benefits, which are only effective to the extent that they are adequately generous and fully funded. 23 And finally, the experience of the Great Recession suggests that even a severe recession need not have a catastrophic impact on family poverty if the safety net is designed to mitigate those effects. Expansion of the eligibility and generosity of both unemployment compensation and SNAP played a particularly important role during the Great Recession. These successful responses to the Great Recession could be made automatic in periods of rising unemployment, especially if proposals to restrict eligibility reduce the availability of safety net programs during periods of low unemployment.
Our analysis shows the benefits of more accurate poverty measures in assessing the size and character of important metrics like the Black–White poverty differential among families with children. The SPMa measurement shows that progress in reducing that differential has been substantial, and that this progress is not doomed to reversal every time there is a recession. The differential is still substantial, however, and reducing it requires that we learn from the past and create a tax and transfer system that goes beyond supporting work and reducing the impact of recessions by treating poverty reduction as a goal, not just as a hoped-for consequence of low unemployment rates. 24
Supplemental Material
sj-docx-1-rbp-10.1177_00346446211006152 - Supplemental material for Family Poverty in Black and White: Results From a New Poverty Measure
Supplemental material, sj-docx-1-rbp-10.1177_00346446211006152 for Family Poverty in Black and White: Results From a New Poverty Measure by Dennis H. Sullivan and Andrea L. Ziegert in Review of Black Political Economy
Footnotes
Acknowledgments
The authors wish to thank Chris Wimer and Erin Todd Bronchetti for their assistance in helping us to compose the dataset, and Ron Oaxaca and Bill Even for econometric counsel.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Funding
The authors received no financial support for the research, authorship and/or publication of this article.
Supplemental material
Supplemental material for this article is available online.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
