Understanding Multidimensional Poverty and its Clustering: A Focus on Race in America, Study notes of Human Development

Download Understanding Multidimensional Poverty and its Clustering: A Focus on Race in America and more Study notes Human Development in PDF only on Docsity!

Richard Reeves, Edward Rodrigue, and Elizabeth Kneebone The Brookings Institution April 2016

FIVE EVILS: MULTIDIMENSIONAL

POVERTY AND RACE IN AMERICA

FIVE EVILS: MULTIDIMENSIONAL POVERTY AND RACE IN AMERICA 2

Introduction

In 1942, at the height of the Second World War, the British academic and former civil servant William Beveridge issued a report titled Social Insurance and Allied Services (1942). Already preparing for peace, Beveridge identified “Five Giant Evils” that needed to be confronted and defeated once the war was won. These five evils were “squalor, ignorance, want, idleness, and disease.” Beveridge believed that all five had to be addressed through concerted government action, with improved housing (“squalor”), universal secondary education (“ignorance”), income transfers to the poor (“want”), full employment (“idleness”), and a national health service (“disease”). Sales of the full Beveridge report broke 100,000 within a month. When a more accessible summary was produced, a further 600,000 copies were distributed. Beveridge, a soft-spoken academic, became a household name. His plan became the animating vision for post- war British society. Although a Liberal,^1 Beveridge helped prepare the ground for the Labour Party’s victory in 1945 and the resulting creation of the National Health Service, universal school system, and social insurance schemes for the unemployed and elderly. Beveridge’s report was not only about poverty in the narrow sense of lack of income, or “want,” but also about poverty and disadvantage as broader concepts. He understood, in other words, that disadvantage is multidimensional. This insight remains a useful one. There is a continuing, mostly facile debate over whether the U.S. won or lost the War on Poverty declared by President Lyndon B. Johnson more than five decades ago. But among its other problems, this argument is often restricted to a narrow, income-based conception of what it means to be poor. Of course poverty is about a lack of money. But it is not only about that. This is one reason many other labels are used: disadvantaged, vulnerable, at-risk, low-skilled, economically insecure, socially excluded, and so on. Poverty as a lived experience is often characterized not just by low income, but by ill health, insecurity, discomfort, isolation, and lack of agency.^2 In practice, of course, the various dimensions of poverty often go together. A lack of paid work almost always means a low income, which can induce stress that leads to health problems, make accessing health care more difficult, and so on.

FIVE EVILS: MULTIDIMENSIONAL POVERTY AND RACE IN AMERICA 4 handful of other specific attempts to construct a multidimensional poverty measure in the U.S. (see Appendix A for a table showing the dimensions, specific indicators, and data sources used in five previous studies). Koohi-Kamali and Liu (2014), who restrict their analysis to Pennsylvania, find high rates of multidimensional poverty among black and Hispanic single mother households. Wagle (2008) differentiates between three broad categories of poverty: what he labels “economic wellbeing poor,” “capability poor,” and “social inclusion poor.” He finds that the risk of being “deeply poor” (i.e. disadvantaged on at least two of the three) or “abject poor” (all three) is much greater for black, Hispanic, and Native American respondents. As he concludes: The multidimensional approach…does not just assess poverty status. It assesses the state of human well-being by focusing on ‘what one has,’ ‘how much prospect one has,’ and ‘how much advantaged or disadvantaged one is in society.’ Dhongde and Haveman (2014) also found significant variations in multidimensional poverty by race; Asian residents suffered from multidimensional disadvantage most frequently, partly because the authors included indicators for “crowded housing” and “lack of English fluency.” Scholars studying multidimensional disadvantage lean heavily on the work of researchers in the human development field like Sabina Alkire and James Foster (2007). The multidimensional approach has been influential in a number of countries (OPHI 2014), but so far has received less attention in more advanced economies. This is unfortunate, since there is growing dissatisfaction with traditional, narrowly income-based measures in many nations, including the U.S. and the UK. There is a danger, however, of going too far the other way, and casting the net too wide. Interpreting a long list of indicators can be difficult.

Five dimensions of poverty

We attempt to steer a middle course between narrowness and complexity and adopt five dimensions of poverty using the 2014 American Community Survey Public Use Microdata Sample (ACS PUMS). Our dimensions and thresholds are as follows:

1. LOW HOUSEHOLD INCOME While poverty is not just about income, income is still important (a lesson lost on the UK government, incidentally, but that’s another story^4 ). For our purposes, respondents are considered poor in terms of income if they are in a household below 150 percent of the federal poverty line (FPL). Why 150 percent of FPL rather than the FPL? Because the FPL is too low—in 2015, $24, for a family of four.^5 When it was set in the 1960s, the FPL was close to 50 percent of median

5 REEVES, RODRIGUE, AND KNEEBONE ECONOMIC STUDIES AT BROOKINGS income. Today, because it has only been adjusted for inflation, it is closer to 30 percent of the median (Smeeding et al. 2011).

2. LIMITED EDUCATION Lack of education inhibits life chances, earning opportunities, and economic security. In the modern labor market, for example, people without a high school diploma are typically at a sharp disadvantage. We therefore adopt this threshold for our analysis. We also include those with GEDs as disadvantaged, since these appear to be less valuable than traditional diplomas in the labor market (Heckman and Rubinstein 2001; Heckman, Humphries, and Kautz 2014). 3. NO HEALTH INSURANCE Ideally, we would construct a measure of ill health as one of our dimensions of disadvantage. The ACS contains questions about disability status, such as blindness, deafness, self-care difficulty, and ambulatory difficulty.^6 But we define a lack of health insurance, either public or private, as our third dimension of disadvantage. This is for two reasons. The first is that disability is potentially subjective; it could also omit other forms of ill health, like diabetes, asthma, hypertension, or high blood pressure. As a binary measure, health insurance coverage is also more similar to our other dimensions. And insurance status captures many aspects of health-related disadvantage that we want to capture. Lacking insurance exposes people to greater health and financial risks in the event of illness. Research also suggests that the uncertainty associated with uninsurance creates ongoing psychological stress for families.^7 4. LOW-INCOME AREA Living in a high-poverty area puts people at a disadvantage, above and beyond their own household’s income-poverty status, because of local factors like the quality of schools, social capital, job connections, and crime.^8 For the purpose of our multidimensional measure, we define disadvantage as living within a Public Use Microdata Area (PUMA)^9 where poverty exceeds 20 percent (here using the standard FPL). PUMAs are statistical geographies created by the Census Bureau. Each contains roughly 100,000 people. In dense New York City, PUMAs are about the size of zip codes; in Dallas, PUMAs encompass three or four zip codes; fewer than 10 PUMAs cover all of sparsely-populated South Dakota. 5. UNEMPLOYMENT Employment brings advantages above and beyond current income, including the prospect of a higher income in the future and a sense of purpose and structure. Of course not all adults need to have a job—especially in a household with caring responsibilities—but it is better to be in a working family than a jobless family, even apart from the obvious economic implications. Our respondents are therefore considered disadvantaged if no one in their household between 25 and 61 is employed.

7 REEVES, R O DRIGU E , AND KNEE BONE ECONOMIC STUDIES AT BROOKINGS The proportion of the population who are disadvantaged on all five dimensions is so small— less than 1 percent—that we do not report results for this group. On the face of it, there is some encouraging news here. While disadvantages do cluster together, a relatively small proportion of overall population suffers from more than two disadvantages at the same time.

Large race gaps in multidimensional poverty

rates

But there may be different risks of multidimensional poverty for different groups or different geographical areas. In what follows, we examine racial differences in the extent to which the dimensions of disadvantage cluster together. There are marked differences in multidimensional poverty rates and patterns by race. Most blacks and Hispanics are disadvantaged on at least one dimension; most whites are not.^13 (We do not report results for Asian Americans here, but they are almost identical to those for whites).^14 Most whites who are disadvantaged on one dimension are not disadvantaged on any others. By contrast, most African Americans and Hispanics who are disadvantaged on one dimension are also disadvantaged on at least one more. 47% 23% 9% 2% 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% at least 1 at least 2 at least 3 at least 4 Percent of population Number of disadvantages

Figure 2. Half face at least one disadvantage

Source: Author's tabulations of 2014 ACS 1-year estimates

FIVE EVILS: MULTIDIMENSIONAL POVERTY AND RACE IN AMERICA 8 Multidimensional poverty, then, is clearly much more common among blacks and Hispanics. While the percentage of all groups with many disadvantages is obviously low, the absolute numbers are not trivial; more than 3 million black and 5 million Hispanic adults suffer from at least three disadvantages. A different way to illustrate this stark race gap is in terms of the relative risk for African Americans and Hispanics of being disadvantaged on multiple dimensions compared to whites. 0% 10% 20% 30% 40% 50% 60% 70% 80% at least 1 at least 2 at least 3 at least 4 Percent of population Number of disadvantages

Figure 3. Blacks and Hispanics face more

disadvantages

White Hispanic Black Source: Author's tabulations of 2014 ACS 1-year estimates 0 1 2 3 4 5 6 at least 1 at least 2 at least 3 at least 4 Ratio Number of disadvantages

Figure 4. The more dimensions of poverty,

the bigger the race gap

White- Black risk ratio White- Hispanic risk ratio Source: Author's tabulations of 2014 ACS 1-year estimates

FIVE EVILS: MULTIDIMENSIONAL POVERTY AND RACE IN AMERICA 10 disadvantages, 71 percent are low-income. For Hispanic adults, the figure is 66 percent, and for black adults, it’s 75 percent. In all the analyses that follow, we adopt this “income-plus” approach to the creation of clusters of disadvantage. Whites are less likely than the other two demographic categories to have both a low household income and some other disadvantage. But there are also clear differences between Hispanic and black residents. Hispanics, for example, have about the same rate of the “low income plus unemployment” disadvantage as whites (both around 6 percent), but four times the risk of having the “low income and no high school diploma” disadvantage compared to whites (17 percent vs. 4 percent). Blacks adults, however, are much more likely than white adults to have the double disadvantage of low income and joblessness, or low income and concentrated geographic poverty. Next we calculate how many people suffer from a combination of at least three disadvantages. The overall rate is of course lower—about 9 percent. Among that 9 percent, the vast majority suffer from the low-income disadvantage. And the race gaps are even larger. Again, black and Hispanic residents suffer from different forms of clustering. Hispanics are almost 10 times more likely than whites to be low-income, without a high school degree, and uninsured (9.6 percent vs. 1.2 percent). On the other hand, black adults are 7 times more likely than white adults to be low-income, live in a high-poverty area, and reside in a jobless household (7.4 percent vs. 1.3 percent). 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% low income and lack of education low income and lack health insurance low income and poor area low income and jobless family Percent of selected population Figure 6. Two disadvantages including low income: Race gaps White Hispanic Black Source: Author's tabulations of 2014 ACS 1-year estimates

11 REEVES, R O DRIGU E , AND KNEE BONE ECONOMIC STUDIES AT BROOKINGS Turning last to the small number of deeply disadvantaged people, those below our disadvantage thresholds on four or even all five categories, the same story emerges. There are almost no white adults in this category. Low-income Hispanics are most at risk of additionally being without health insurance, having less than a high school education, and living in a poor area. For black Americans, being in a jobless household is a bigger risk factor. 0% 1% 2% 3% 4% 5% 6% 7% 8% 9% 10% low income, lack of education, and health insurance low income, lack of education, and poor area low income, lack of health insurance, and poor area low income, lack of education, and jobless family low income, lack of health insurance, and jobless family low income, poor area, and jobless family Percent of selected population

Figure 7. Three disadvantages including

low income: Race gaps

White Hispanic Black Source: Author's tabulations of 2014 ACS 1-year estimates 0% 1% 2% 3% 4% low income, lack of education and health insurance, and poor area low income, lack of education, jobless family, and poor area low income, lack of health insurance, jobless family, and poor area low income, lack of education and health insurance, and jobless family Percent of selected population

Figure 8. Four disadvantages including

low income: Race gaps

White Hispanic Black Source: Author's tabulations of 2014 ACS 1-year estimates

13 REEVES, R O DRIGU E , AND KNEE BONE ECONOMIC STUDIES AT BROOKINGS If anything, the gap between Hispanics and the rest of the population in terms of health insurance coverage has widened in the last 20 years.^16 This suggests that the clustering of low-income status and lack of health insurance has increased. However, the trend among adults has at least stabilized in the last few years. Among Hispanic children the picture is rosier, with a drop from 16 to 10 percentage points in the portion of the population that was uninsured between 2009 and 2014, according to research by La Raza and the Georgetown University Health Policy Institute (Schwartz et al. 2016). If the ACA has the intended effect of expanding coverage, there ought to be a slow de- clustering of these two disadvantages in the years to come. On balance, then, we might expect that Hispanic multidimensional disadvantage will abate to some degree. However, that might not be the case for black adults. First, the black/white employment gap has shown little sign of improvement, especially for men: their black/white employment gap has remained between 13 and 18 percentage points over the last 20 years. 0% 10% 20% 30% 40% 50% 60% 1993 1998 2003 2008 2013 Percent of 25-35 year-olds uninsured

Figure 10. Health insurance coverage: Race gaps

Hispanic Black non- Hispanic White non- Hispanic Source: Author's tabulations of IPUMS CPS; University of Minnesota

FIVE EVILS: MULTIDIMENSIONAL POVERTY AND RACE IN AMERICA 14 Second, the risk of living in a poor area remains significantly higher for black families. In the 1990s there was some improvement on this front, but according to recent work by Elizabeth Kneebone and Natalie Holmes (2016), the recession brought that progress to an abrupt halt. Between the 2000 decennial census and the 2010-2014 American Community Surveys, the chances that black Americans living below the FPL in the nation’s 100 largest metro areas also resided in an extremely poor census tract (where more than 40 percent of residents lived below the poverty line) rose from 1 in 5 to more than 1 in 4. 0% 5% 10% 15% 20% 25% 30% 35% 40% 1993 1998 2003 2008 2013 Percent of non -military males age 25-35 unemployed

Figure 11. Joblessness: Stubborn black-white gaps

Black non- Hispanic men Hispanic men White non- Hispanic men Source: Author's tabulations of IPUMS Current Population Survey; University of Minnesota 20% 26% 14% 18% 3% 6% 0% 5% 10% 15% 20% 25% 30% 2000 2005-2009 2010- Percent of poor population in census tracts with poverty rates 40% or higher in 100 largest metro areas

Figure 12. High-poverty neighborhoods: Race gaps

Black Hispanic Non- Hispanic White Source: Elizabeth Kneebone and Natalie Holmes' tabulations of decennial census and American Community Survey data

FIVE EVILS: MULTIDIMENSIONAL POVERTY AND RACE IN AMERICA 16

Endnotes

1. “Liberal” here refers to the UK Liberal Party, one of the two major parties in the 19th and early 20th centuries. See https://en.wikipedia.org/wiki/Liberal_Party_(UK).
2. Previous analyses of the “underclass” in the U.S. provide examples of geographically-based multidimensional frameworks. See Wilson (1987) or Sawhill and Jargowsky (2006).
3. Child opportunity maps for U.S. metropolitan areas can be found here: http://www. diversitydatakids.org/data/childopportunitymap.
4. For commentary, see Reeves 2015.
5. The 2016 federal poverty guidelines can be found here: https://aspe.hhs.gov/poverty- guidelines.
6. See the link below for a list of the variables in the 2014 ACS public-use microdata: http://www2.census.gov/programs-surveys/acs/tech_docs/pums/data_dict/PUMSDataDict14.pdf
7. For recent evidence, see Finkelstein et al. 2012.
8. For a detailed analysis of concentrated poverty, see Kneebone and Holmes 2016.
9. For more information on PUMAs, see https://www.census.gov/geo/reference/puma. html.
10. The ACS includes a question about the length of time since respondents last worked, but the possible responses are somewhat broad: “within the past 12 months,” “1 to 5 years ago,” or “over 5 years ago or never worked.” See here for a copy of the 2014 ACS questionnaire: https:// www.census.gov/programs-surveys/acs/methodology/questionnaire-archive.2014.html.
11. Excluding active duty members of the military and people living in group quarters, like college dormitories, nursing homes, or correctional facilities.
12. Part of the character of “deeper” multidimensional disadvantage appears in the income figures. Adults with at least 3 disadvantages have household incomes that average only 93 percent of the federal poverty line. Those with one or more disadvantages average 220 percent of the FPL.
13. In constructing our racial categories, we have followed the approach of William H. Frey and others. Respondents are categorized as “white” and “black” based on their own definition,

17 REEVES, R O DRIGU E , AND KNEE BONE ECONOMIC STUDIES AT BROOKINGS but only if they described themselves as “non-Hispanic.” “Hispanics” include those who defined themselves as such, as well as in many cases describing themselves as “white” or “black.” See Frey 2014. Our sample also includes non-citizens.

14. Citizenship status makes a difference; re-tabulating the results while omitting non- citizens lowers Hispanic rates of disadvantage. The new levels are generally comparable to those experienced by black adults. 61 percent of Hispanic citizens face at least one disadvantage (versus 71 percent of all Hispanics), 31 percent face two or more (versus 43 percent), 13 percent face three or more (versus 21 percent), and 4 percent face four or more (versus 6 percent). We choose to include non-citizens in our main results, since they still participate in most aspects of American life through their workplaces and communities. (Only about a third of immigrants are unauthorized, according to analyses by the Pew Research Center. See Passel, Cohn, and Gonzalez-Barrera 2013.)
15. There’s some debate about how much of this change represents real improvement, and how much resulted from lowering graduation standards. See Kamenetz 2015 for more.
16. Here, too, the citizenship status of the Hispanic population makes a difference. Roughly 19 percent of Hispanic citizens lack health insurance (versus 33 percent in the tabulations above).
17. See Reeves, “Two anti-poverty strategies” for more.

19 REEVES, R O DRIGU E , AND KNEE BONE ECONOMIC STUDIES AT BROOKINGS Author/ Date Dimensions Indicators Data Source(s) Selected conclusions Dhongde and Haveman (2014)

• Health
• Education
• Standard of living
• Housing
  • Health insurance coverage
  • Disability status
  • High school completion
  • English proficiency
  • Poverty status
  • Employment status
  • Whether there are more occupants than rooms in a home
  • Housing costs exceed 50% of income 2011 American Community Survey (ACS) Public Use Microdata Sample - In 2011, one in five adults was multidimensional poor; compared to an official poverty estimate of 13%. Multidimensional poor experienced about 7% of all deprivations possible. - Variation explained more by race, nativity, and region than by age or gender. Koohi- Kamali and Liu (2014)
• Education
• Work
• Income
• Standard of living
• High school completion
• Employment status of household head and spouse
• SNAP benefits
• Public assistance income
• SSI income
• More than 2 residents per bedroom
• Vehicle ownership
• Real estate ownership 2006-2010 ACS Public Use Microdata Sample (specifically for Pennsylvania)
• Hispanics most deprived in educational dimension; blacks most deprived in employment dimension.
• For full sample, most significant dimension of multidimensional poverty is work status (contributes 41% of total deprivation.) Wagle (2008)
• Economic well-being
• Capability
• Social inclusion
• Respondent income
• Total family income
• Satisfaction with financial situation
• Educational attainment
• Health condition
• Feel as though people are treated with respect at work
• Occupational prestige
• Industry
• Work status
• Weeks of work
• Self-employment indicator
• Activism
• Voted in 2000
• Group membership
• Associational activity, and perceived importance of associational activities
• Number of friends and relatives 2004 General Social Survey
• The Northeast has lower multidimensional poverty rates; the South has the highest rates.
• Blacks, Hispanics, and American Indians suffer from multidimensional poverty at disproportionate rates. The same is true for widowed Americans.

FIVE EVILS: MULTIDIMENSIONAL POVERTY AND RACE IN AMERICA 20

References

Acevdeo-Garcia, Dolores, Nancy McArdle, Erin F. Hardy, Unda Ioana Crisan, Bethany Romano, David Norris, Mikyung Baek, and Jason Reece. “The Child Opportunity Index: Improving Collaboration Between Community Development and Public Health.” Health Affairs 33 (2014): 1948-1957. Alkire, Sabina, and James Foster. “Counting and Multidimensional Poverty Measurement.” Oxford Poverty & Human Development Initiative Working Paper Series no. 7, Oxford, UK, January

2008. http://www.ophi.org.uk/wp-content/uploads/ophi-wp7.pdf. Beveridge, Sir William. “Social Insurance and Allied Services.” Paper Presented to UK Parliament, London, UK, November 1942. Burtless, Gary. “Raising everyone’s retirement age undercuts a key goal of Social Security.” The Dallas Morning News , October 22, 2015. Dhongde, Shatakshee, and Robert Haveman. “Multi-dimensional Poverty in the U.S.” Working Paper, Institute for Research on Poverty, University of Wisconsin, Madison, WI, April 2014. http://www.irp.wisc.edu/newsevents/seminars/Presentations/2013-2014/US_MPI_dhongde_ haveman.pdf. Finkelstein, Amy, Sarah Taubman, Bill Wright, Mira Bernstein, Jonathan Gruber, Joseph Newhouse, Heidi Allen, Katherine Baicker, and the Oregon Health Study Group. “The Oregon Health Insurance Experiment: Evidence from the First Year.” The Quarterly Journal of Economics 127 (2012): 1057-1106. Frey, William H. Diversity Explosion: How New Racial Demographics are Remaking America. Washington, DC: Brookings Institution Press, 2014. Heckman, James J., John Eric Humphries, and Tim Kautz, eds. The Myth of Achievement Tests: The GED and the Role of Character in American Life. Chicago, IL: University of Chicago Press,

Heckman, James J., and Yona Rubinstein. “The Importance of Noncognitive Skills: Lessons from the GED Testing Program.” American Economic Review 91 (2001): 145-149. Kamenetz, Anya. “High School Graduation Rates: The Good, the Bad, and the Ambiguous.” Washington, DC: National Public Radio, June 9, 2015. https://www.census.gov/geo/reference/ puma.html. Kneebone, Elizabeth, and Natalie Holmes. “U.S. concentrated poverty in the wake of the Great Recession.” Washington, DC: Brookings Institution, 2015. http://www.brookings.edu/research/ reports2/2016/03/31-concentrated-poverty-recession-kneebone-holmes. Koohi-Kamali, Feridoon, and Ran Liu. “U.S. Multidimensional Poverty by Race and Motherhood: Evidence from Pennsylvania Census Data.” Working Paper, Emory College of Arts and Sciences, Atlanta, GA, 2014. http://economics.emory.edu/home/documents/workingpapers/