La Follette Policy Report Volume 25, Number 1, Fall 2015 Director’s Perspective Poverty Symposium Showcases Longtime Milwaukee Partnership In our long tradition of partnering on research and scholarship in the city of Milwaukee, the La Follette School held a half-day conference on “Urban Men in Poverty: Problems and Solutions” in April in partnership with the Marquette University Law School. La Follette School economist Geoffrey Wallace organized the conference, and he kicked it off with extensive insights into facts and figures on problems and solutions related to urban poverty, including reasons for focusing on men and on urban centers in studying these issues. Geoffrey described “a mass retreat from employment for less skilled men” in recent times, a trend that is particularly strong among African American men. You can read more about Geoffrey’s observations in this issue of the La Follette Policy Report. Professor Charles Franklin of Marquette University Law School presented data on the high percentage of young men who are arrested and incarcerated by their mid-30s and how that harms their Director’s Perspective continued on page 15 Inside this Issue The Measurement of Poverty: An Evolving Story..............................................8 Racial Variation in the Effect of Incarceration on Neighborhood Attainment..........................11 The Effect of Child Support on the Work and Earnings of Custodial Mothers: New Evidence...........................................................16 Adult Male Poverty is Cause for Concern By Geoffrey Wallace P oor working-aged men tend to be ignored in discussions of poverty among policymakers, academics, and the popular press. Part of the reason for this lapse is that their poverty rates are lower than those of other groups; these men are also dismissed as being “undeserving” poor. Indeed, the official poverty rates for nonelderly and elderly men are below those of children, nonelderly adult women, and elderly adult women. However, the government’s supplemental poverty measure that incorporates noncash government benefits and refundable tax credits finds poverty rates for adult men that are in line with those of other demographic groups. The government directs noncash refundable tax credits and noncash transfers to families with children, which explains higher male poverty relative to other groups in the supplemental poverty measure. Figure 1 shows the 2013 official and supplemental poverty rates for children, adults 19 to 64, and older adults. The supplemental measure indicates a greater share of men live in poverty: 14.8 percent of men ages 19-64 and 11.9 percent of men 65 and older, compared to official rates of 11.7 percent and 6.8 percent. Reducing male poverty should be more of policy concern for two reasons. First, working-age male poverty rates vary substantially by demographic group, with some groups experiencing a high incidence of poverty; moreover, poverty for this group has far-reaching generational repercussions that affect children. Male poverty affects child poverty. In 2013, about one-third of men whom the supplemental poverty measure deemed poor lived in households with at least one child. More than 4 percent of poor nonelderly men who do not live with children are responsible for paying support for noncustodial children. Reducing the incidence of male poverty by increasing income, reducing net tax liability, or reducing out-of-pocket medical expenses could lead to substantial reductions in child poverty. For example, eliminating out-of-pocket medical expenses for poor nonelderly adult men or increasing their income by 10 percent would reduce the supplemental measure’s child poverty rate by more than a percentage point. Incidence and Composition of Male Poverty in 2013 Table 1 shows the supplemental poverty measure rates in 2013. The numbers in the table were obtained by matching demographic information from the 2014 Current Population Survey Annual Social and Economic Supplement (CPS) to the 2013 Supplemental Poverty Measure Research File. As Table 1 shows, the incidence of poverty varies dramatically across subgroups. Among men defined as poor by the Figure 1. Poverty Rates by Age, Group, and Sex, 2013 22% Poverty Rates According to Official Measure 20.2% 20% 18% Poverty Rates According to Supplemental Measure 16.8% 16.7% Percentage of Poor People 16% 15.2% 14.8% 15.8% 14% 11.9% 11.7% 12% 11.6% 10% 8% 6.8% 6% 4% 2% 0 Children Ages 0 to 18 years Men Ages 19 to 64 Years Women Ages 19 to 64 Years Men 65 Years and Older Women 65 Years and Older Source: Authors’ calculations supplemental poverty measure, those ages 19 to 24 have a high incidence of poverty at 22.4 percent poor—a rate higher than the child poverty rate of 16.7 percent. Poverty rates for men older than 25 are relatively low. This situation is especially true for men older than 65. For men ages 19 to 64, race and location are important determinants of poverty status as measured by the supplemental measure and shown in Table 1. Male poverty rates among blacks and Hispanics are 22.3 and 24.6 percent respectively, compared to 10.5 percent for non-Hispanic whites and 16.1 percent for other racial-ethnic groups. The male poverty rate in the western United States is 3 percentage points higher than poverty rates in the South, the next highest poverty rate region. The high rate of western male poverty can be attributed to the relatively high cost of shelter in the West. Controlling for household size and composition and for metropolitan status, the poverty thresholds (the minimum money a household needs to not be living in poverty) applicable to Geoffrey Wallace is an associate professor of public affairs and economics at the La Follette School. His research is in labor economics, the economics of marriage and the family, and policy issues relating to poverty. This article is based on a presentation he gave in April at the La Follette School conference on “Urban Men in Poverty: Problems and Solutions” held in partnership with the Marquette University Law School. 2 / La Follette Policy Report poor men in the West are more than $1,000 higher than the poverty thresholds applicable to men in the Northeast, the region with the next most expensive shelter costs. The poverty rate in metro areas, particularly in principal cities, is higher than the poverty rate in non-metro areas. In the principal cities identified in the survey, the nonelderly adult male poverty rate is 19.3 percent compared to 15.2 percent for metro areas as a whole (which includes principal cities) and 12.5 percent for rural areas. Rates of poverty for nonelderly men in families (including cohabitating arrangements) are fairly low. Moreover, nonelderly men living with children have a poverty rate that is lower than the average rate across the entire male nonelderly adult population. Men in married-couple families face the lowest risk of poverty (9.4 percent) followed by men who reside in resource-sharing units headed by a cohabitating couple (14.9 percent). Both of these groups experienced less than the average incidence of poverty in 2013. Men living in other types of family and non-family arrangements have a high incidence of poverty: Poverty rates for men in male-headed families, femaleheaded families, and non-families are 21.6 percent, 25.4 percent, and 25.7 percent, respectively. Education level and worker status have a large influence on the incidence of poverty. Nearly one-third of nonelderly adult men without a high school education are poor. Men with just a high school education and no college have a poverty rate of 17.4, while men with a college degree or more have a poverty www.lafollette.wisc.edu Fall 2015 rate of 6.3 percent. Nonelderly adult men who reported to the CPS that they did not work at jobs or for businesses in 2013 have a poverty rate of 36.6 percent. Men who reported working part time or for less than a full year experienced a poverty rate of 22.5 percent. Nonelderly adult men who work full time all year round have a very low incidence of poverty. The last column of Table 1 shows the percentage of the nonelderly adult poor men within each subgroup. A subgroup can comprise high share of poor men because they make up a large share of the population, because they have a high rate of poverty, or both. Groups falling into the first category include white males (45.1 percent of nonelderly poor adult men), males living in metropolitan areas, males living in families with or without children, men in married-couple families, men with some college but no degree, and men who worked full time year round. Men who make up a large share of the poor because of high poverty rates include men ages 19 to 24 and men not living in families. Groups of men who represent a large share of the poor because of high poverty rates and because they make up a large share of the population include black and Hispanic men, men residing in the West, men with a high school education or less, and men who worked less than full time and year round. The connections among poverty, education, and worker status are striking. More than 60 percent of nonelderly poor adult men have a high school diploma or less education, and more than 75 percent of poor men work less than full time and year round. Almost 43 percent of nonelderly adult poor men have a high school diploma or less education and work less than full time year round, and nearly 60 percent of these poor less educated and un- and underemployed men are black or Hispanic. Why Aren’t More Poor Men Working? Responses in the CPS give some insight into the reasons men reported for not working in a calendar year, including being ill or disabled, taking care of home or family, going to school, not being able to find work, or other reasons. This information provides evidence of significant barriers to employment for the approximately 50 percent of poor men who did not work in 2013. More than 35 percent of nonelderly poor adult men who did not work in 2013 listed illness and disability as the primary reason. Another 10 percent said they were taking care of home or family, which might suggest a relative’s health problems. Nearly 18 percent of poor men who did not work in 2013 reported difficulty finding a job as the primary reason. There is considerable heterogeneity in reported reasons for not working across age groups. Within the group of poor men ages 19 to 24 who did not work in 2013, only 8.3 percent reported not working because of illness or disability, and 60 percent reported not working because they were enrolled in school. Among nonelderly poor men 45 and older who did not work in 2013, more than 51.5 percent reported disability or illness as the primary reason, with another 21 percent reporting that they were taking care of home or family. For this group of older men, the inability to find a job does not appear to be an Fall 2015 Table 1: Male Poverty per the Supplemental Poverty Measure, 2013 Percentage Percentage of Poor Men within Subgroup within Subgroups that is Poor Subgroup By Age 19-24 years 25-34 years 35-44 years 45-54 years 55-64 years 65+ years 22.4 15.2 13.0 12.4 13.3 11.9 18.2 19.9 15.7 16.0 15.7 14.5 By Race/Ethnicity for Ages 19-64 White (non-Hispanic) 10.5 Black (non-Hispanic) 22.3 Hispanic 24.6 Other 16.1 45.1 17.9 28.9 8.2 By Region for Ages 19-64 Northeast Midwest South West 13.215.4 12.216.9 14.9 33.5 17.934.2 By Metro for Ages 19-64 Metro Non-metro Not identifiable 15.2 87.7 12.511.8 10.8 0.6 By Family for Ages 19-64 Non-family Family without children Family with children 25.730.3 12.5 37.5 12.4 32.2 By Head of Family Type for Ages 19-64 Married couple 9.4 Cohabitating couple 14.9 Male-headed family 21.6 Female-headed family 25.4 Male nonfamily 25.7 37.9 9.0 11.4 11.3 30.3 By Education Level for Ages 19-64 Less than high school 32.2 24.9 High school, no college 17.4 36.4 Some college 13.4 26.4 College+ 6.312.3 By Employment Status for Ages 19-64 Did not work 36.6 Did not work full time or year around 22.5 Worked full time year round 5.6 46.5 30.5 23.1 Source: 2014 Current Population Survey Annual Social and Economic Supplement to the 2013 Supplemental Poverty Measure Research File www.lafollette.wisc.edu La Follette Policy Report / 3 important factor in explaining the lack of employment in 2013, with less than 15 percent indicating that they did not work because of difficulty to finding work. Black and Hispanic men of all ages are more likely to report not working due to difficulty finding work. Poor Men and Sources of Income Table 2: Income Sources and Deductible Expenses of Poor Men Ages 19 to 64 not in Families or Cohabitating Relationships Average Amount Across All Nonelderly Men Net Individual Resources Total Cash Income Total earnings Own earnings Other earnings Unemployment Insurance, welfare, Supplemental Security Income Social Security Other cash transfers Total Noncash Transfers Total Tax Liability Federal tax liability before credits Federal tax credits Payroll taxes State taxes Total Deductible Expenses Child support Work expenses Out-of-pocket medical expenses $4,795 7,244 5,046 5,046 0 Percentage with Income from Source Average 78.3 79.4 52.9 52.9 0 $6,919 9,184 9,540 9,540 0 Nearly 50 percent of poor men did not work in 2013. Of those who 699 12.2 5,742 did work, most worked less than 690 7.9 8,683 full time year round. This finding 809 27.0 2,997 suggests that poor men have mod611 25.2 2,421 est earnings, which prompts several -552 46.2 -1,201 questions: What other sources of -146 22.5 -648 income do these men have? What 77 26.5 290 is their tax liability? To what extent -421 51.2 -822 do they incur expenses such child -63 17.8 -375 support payments, work expenses, -2,508 86.2 2,909 and out-of-pocket medical expens-348 5.8 -6,007 es? The following analysis addresses -653 52.3 -1,248 these questions by breaking non-1,508 75.3 -2,002 elderly poor men into rough three equal groups based on their livingPoverty Threshold 11,787 family arrangements and examining income sources and expenses of Poverty Gap 6,992 each group. As tables 2-4 show, the groups are nonelderly adult men Notes: Numbers may not add due to rounding or to state tax credits. Also, an individual may have negative net who are not in families (or cohabi- resources. Source: 2014 Current Population Survey Annual Social and Economic Supplement to the 2013 Supplemental tating), nonelderly adult men in Poverty Measure Research File families without children, and nonelderly adult men in families with children. The tables are Table 2 provides information on income sources available organized in a way that is consistent with the calculation of the to poor men who are not in a family or cohabitating relationsupplemental poverty measure. ship. On average for a man not in a family or cohabitating The supplemental poverty measure defines net individual relationship, a poverty gap of $6,992 exists between net indior family resources as the sum of earnings, cash transfers, nonvidual resources of $4,795 and the supplemental poverty meacash transfers, net tax liability, and deductible expenses. These sure poverty threshold of $11,787. income sources appear in the top panel of each table. DeductAlthough 79.4 percent of poor working-age non-family ible expenses include work expenses, child support payments, men have some net cash income, average levels are quite low, and out-of-pocket medical expenses and are shown in the botaveraging $7,244. The biggest source of cash income is earntom panel of each table. Net individual or family resources are ings, which 52.9 percent of men report receiving, an average of compared to the supplemental poverty thresholds to determine $5,046. Rates of receipt of Unemployment Insurance, welfare, an individual’s or sharing unit’s poverty status. The poverty gap and Supplemental Security Income are low, but these income is the amount of additional resources the individual or famsources are important for the men who rely on them. Traditionily would need in order not to be poor—to reach the poverty ally, few people in the 19-to-64 age range collect Social Security, threshold. In each of these three tables, the first column shows but the 7.9 percent of men who do receive Social Security averthe average amount associated with the income or expense catage $8,683 in payments. About a quarter of poor working-age egory across all nonelderly men, the second column shows a men receive noncash transfers and other cash transfers, which positive (income) or negative (expense) percentage of workcan represent anything from educational assistance to financial ing age men with income from each source, and the last colassistance from friends and relatives. umn shows the average of the positive (resources) or negative Poor working-age men by and large do not receive any help (expenses) values. from the tax system, and they have other expenses that decrease 4 / La Follette Policy Report www.lafollette.wisc.edu Fall 2015 welfare, and Supplemental Security Income are low. Evidence that some Table 3: Income Sources and Deductible Expenses of these men live with older relatives of Poor Men Ages 19 to 64 in Families without Children comes in the form of high rates of Social Security receipt relative to Average Amount Percentage nonfamily men. A little over a third Across All with Income of the men live in resource-sharing Nonelderly Men from Source Average units that receive cash transfers Net Family Resources $12,525 89.9 $14,749 other than Unemployment Insur Total Cash Income 20,049 93.0 21,577 ance, welfare, Supplemental Secu Total earnings 14,533 70.7 20,545 rity Income, and Social Security, Own earnings 5,987 48.8 12,257 and a little over a quarter live in Other earnings 8,546 51.2 16,689 resource-sharing units that receive Unemployment Insurance, welfare, noncash transfers. Similar to poor Supplemental Security Income 1,182 16.4 7,207 nonfamily men, poor family men Social Security 2,748 22.5 12,223 without children pay taxes and do Other cash transfers 1,585 34.6 4,575 not receive much in the way of cred Total Noncash Transfers 927 28.7 3,228 its from the tax systems. Their fami Total Tax Liability -1,697 65.6 -2,642 lies also have substantial work and Federal tax liability before credits -494 40.4 -1,222 out-of-pocket medical expenses. Federal tax credits 201 35.7 564 The income sources of the 32 Payroll taxes -1,214 70.4 -1,725 percent of nonelderly poor men State taxes -191 27.1 -724 who live with children are shown Total Deductible Expenses -6,754 96.0 -7,037 in Table 4. These men have an aver Child support -188 3.0 -6,346 age poverty threshold of $30,173, Work expenses -1,527 70.5 -2,165 net family resources of $19,991, Out-of-pocket medical expenses -5,039 89.1 -5,657 an average poverty gap of $10,182, and income sources that are quite Poverty Threshold 22,799 different from men living in famiPoverty Gap 10,274 lies without children. How do their income sources differ? First of all Notes: Numbers may not add due to rounding or to state tax credits. Also, an individual or a family may have their own earnings are substantially negative net resources. higher than the other two groups of Source: 2014 Current Population Survey Annual Social and Economic Supplement to the 2013 Supplemental men. Second, they are far more likePoverty Measure Research File ly to receive noncash transfers (78 percent do) and when they do, the amounts of these transfers their net resources. About half of the men face some tax liability received are higher. Lastly, their families receive a substantial with an average amount of $552. Because their income levels amount of assistance from the tax system. Although the famiare typically very low, most of this liability comes from paylies of 28 percent of these men do have some net tax liability, roll taxes. The largest source of deductible expenses are outon average they are receiving net benefits from the tax system of-pocket medical expenses that average $1,508 per year and in the amount of $1,062 (largely through the earned income are incurred by three-quarters of nonelderly poor men. For the tax credit). On average, their families have about $2,100 in tax roughly half of men who do work, deductible work expenses liability from federal, state, and payroll taxes, but this liability is average $1,248 per year. more than completely offset by tax credits that average $3,145. The income sources of the 37 percent of nonelderly poor Similar to families without children, families with children men who live with family members (including cohabitating have substantial deductible expenses: $7,600 on average. Outpartners) but do not live with any children are shown in Table of-pocket medical expenditures are similar to families without 3. The average poverty gap for these nonelderly poor men children, but due to child-care cost, work expenses for families is $10,274, with an average threshold of $22,799. Levels of with children are higher. employment and earnings for men in families are similar to those of nonfamily men; about half the poor nonelderly family men not living with children work with average earnings for Conclusions workers at $12,257 per year. About half of these poor men live Male poverty is a problem that historically has not received with family members who earn income. On average, working much attention in academic and policy circles or in the popurelatives provide $8,546 in earnings, a level that is more than lar press. Part of this lack of attention could be attributed to $2,500 greater than the average level of earnings of the males the fact that males have traditionally been thought of as a low themselves. Rates of receipt of Unemployment Insurance, Fall 2015 www.lafollette.wisc.edu La Follette Policy Report / 5 poverty group. While this status is undoubtedly still true for elderly Table 4: Income Sources and Deductible Expenses males, nonelderly adult males have of Poor Men Ages 19 to 64 in Families with Children supplemental poverty measure poverty rates that are very similar Average Amount Percentage to those of children and women. Across All with Income Moreover, nearly one-third of non Nonelderly Men from Source Average elderly poor men living in resourceNet Family Resources $19,991 95.8 $21,382 sharing units with children and Total Cash Income 23,623 95.3 24,818 more than 4 percent of nonelderly Total earnings 19,997 83.9 23,824 poor men who do not live with Own earnings 11,517 66.5 17,318 children are responsible for paying Other earnings 8,480 50.6 16,769 child support for a noncustodial Unemployment Insurance, welfare, child: Male poverty and child pov Supplemental Security Income 1,238 18.1 6,831 erty are related. Social Security 1,261 10.3 12,280 Nonelderly men differ in their Other cash transfers 1,128 28.0 4,031 risk of being poor. Men ages 19 to Total Noncash Transfers 2,733 78.4 3,488 64 are at an elevated risk for pov Total Tax Liability 1,062 28.1 -3,317 erty, including young men, black Federal tax liability before credits -327 25.7 -1,275 men, Hispanic men, and urban Federal tax credits 3,145 71.8 4,380 men, men residing in the West, and Payroll taxes -1,661 83.4 -1,992 men who are not in families or are State taxes -94 22.7 -803 in female-headed families. There Total Deductible Expenses -7,728 97.7 -7,601 are strong links between education Child support -155 2.4 -6,515 level and the risk of poverty, with Work expenses -2,168 83.5 -2,597 those with a high school degree or Out-of-pocket medical expenses -5,105 91.0 -5,610 less education facing a much higher risk of poverty. Employment and Poverty Threshold 30,173 poverty also are related: 5.6 percent of nonelderly men who reported Poverty Gap 10,182 working full time year round in 2013 were poor compared to more Notes: Numbers may not add due to rounding or to state tax credits. Also, an individual or a family may have than 36 percent of men who report- negative net resources. Source: 2014 Current Population Survey Annual Social and Economic Supplement to the 2013 Supplemental ed not working. Poverty Measure Research File The extent to which nonelderly poor men receive assistance from the tax and transfer system is Hispanic men and that being an ex-offender may represent a highly dependent on whether they reside in households with substantial barrier to employment, as a 2003 study by Devah children. Men living in family units without children are highly Pager suggests. Moreover, although not reflected in the data dependent on earnings as a source of resources, while men livsource used to calculate poverty and employment rates in this ing in families with children have higher earnings and are far analysis, child support payments and arrears may discourage more likely to receive noncash transfers (food stamps, housing formal sector work by acting as a large lump-sum tax on earnassistance, energy assistance, etc.) and net benefits from the tax ings of poor men, as a 2012 study by Daniel Miller and Ronald system. Mincy suggests. Looking at the future of male poverty, the news is good A source of potential good news is found in the deductand bad. The bad news first: Male poverty is closely tied to ible expenditures of poor men and their families and the ways education and employment, neither of which is very tractable in which policy is likely to reduce these expenses in the near in the near term. Forty-seven percent of nonelderly men who future. As seen in the analysis of income sources and expenses, were poor in 2013 reported no formal sector employment for the resource-sharing units to which poor men belong face subthe calendar year. While information on the duration of nonstantial out-of-pocket medical expenditures. For nonelderly employment is not available, it seems likely that four years into poor men who are not in families, average out-of-pocket medithe recovery from the Great Recession many of these men have cal expenditures account for more than 20 percent of the povnot worked in recent years and will not work in future years. erty gap. For nonelderly poor men in families (with and withIllness and disability are important reasons for non-employout children) average out-of-pocket medical expenses account ment as is difficulty finding a job, particularly for black and for 50 percent of the poverty gap. Indeed, high out-of-pocket Hispanic men. It seems likely that the ex-offender population medical expenditures may be an important determinant of is heavily represented in this group of poor detached black and male poverty status. Comparing poor and non-poor men with 6 / La Follette Policy Report www.lafollette.wisc.edu Fall 2015 resources less than two times the supplemental measure’s poverty threshold, out-of-pocket medical expenses were $1,000 higher for the families of poor men than for the families of their non-poor counterparts. Not only are out-of-pocket medical expenditures seemingly important in determining male poverty, male poverty is closely associated with not working—and the most frequent reason poor men report not working is illness or disability. A major feature of the Affordable Care Act was the expansion of Medicaid to cover all adults up to 135 percent of the poverty line. At present 30 states and Washington, D.C., have opted to take the Medicaid expansion, one state (Utah) is undecided, and 19 states (mostly in the South and West) have rejected the expansion. Short- and long-term Medicaid expansion has the potential to dramatically improve the prospects of poor men and their families. To help ease male poverty in the short-term, out-of-pocket medical expenses should be substantially reduced for poor men and their families in states that opted to expand Medicaid. In the long-run, Medicaid expansion may lead to increased access to health care, better health outcomes, reduced rates of disability and illness, and higher employment rates. u 1225 Observatory Drive, Madison, WI 53706 policyreport@lafollette.wisc.edu Policy Report Robert Haveman Faculty Editor Karen Faster Senior Editor Faculty Susan Webb Yackee Director, Professor of Public Affairs and Political Science Hilary Shager Associate Director Rebecca M. Blank University Chancellor Menzie Chinn Professor, Public Affairs and Economics Maria Cancian Professor, Public Affairs and Social Work J. Michael Collins Associate Professor of Public Affairs and Human Ecology Mark Copelovitch Associate Professor, Public Affairs and Political Science Jim Doyle Adjunct Associate Professor of Public Affairs Dennis Dresang Professor Emeritus, Public Affairs and Political Science Jason Fletcher Associate Professor of Public Affairs Bob Hanle Adjunct Associate Professor of Public Affairs Robert Haveman Professor Emeritus, Public Affairs and Economics Pamela Herd Professor, Public Affairs and Sociology Karen Holden Professor Emeritus, Public Affairs and Consumer Science Leslie Ann Howard Adjunct Associate Professor of Public Affairs Valerie Kozel Adjunct Associate Professor of Public Affairs Robert J. Lavigna Adjunct Associate Professor of Public Affairs Christopher Lecturer, Public Affairs and Economics McKelvey Robert Meyer Research Professor, Public Affairs Donald Moynihan Professor, Public Affairs Dave Nelson Associate Lecturer of Public Affairs Gregory Nemet Associate Professor, Public Affairs and Environmental Studies Andrew Reschovsky Professor Emeritus, Public Affairs and Applied Economics Timothy Smeeding Professor, Public Affairs Emilia Tjernström Assistant Professor of Public Affairs Geoffrey Wallace Associate Professor, Public Affairs and Economics David Weimer Professor, Public Affairs and Political Science John Witte Professor Emeritus, Public Affairs and Political Science Barbara Wolfe Professor, Public Affairs, Economics, and Population Health Sciences © 2015 Board of Regents of the University of Wisconsin System. The La Follette Policy Report is a semiannual publication of the Robert M. La Follette School of Public Affairs, a teaching and research department of the College of Letters and Science at the University of Wisconsin–Madison. The school takes no stand on policy issues; opinions expressed in these pages reflect the views of individual researchers and authors. The University of Wisconsin-Madison is an equal opportunity and affirmative-action educator and employer. We promote excellence through diversity in all programs. Fall 2015 www.lafollette.wisc.edu La Follette Policy Report / 7 The Measurement of Poverty: An Evolving Story By Rebecca Blank W hen the War on Poverty was launched in 1964, there were no government statistics on the size and composition of the poor population; indeed, there was no accepted definition of poverty or what it meant to be called “poor.” A statistical measure of poverty was needed to indicate how many people were poor, show how the prevalence of poverty was concentrated among different groups, and enable tracking of the poor population over time. Such a measure could also provide a crude indicator of the effectiveness of antipoverty policies. The Creation of a Poverty Measure in the Johnson Administration When the Johnson administration asked the Social Security Administration to propose a poverty definition, Mollie Orshansky was put in charge. She calculated a poverty threshold that presumes the resource-sharing unit to be the family (defined as two or more related individuals residing in the same dwelling); the threshold for families of two or more in 1963 was based on the following definition: poverty threshold = 3 × subsistence food budget The subsistence food budget was the Economy Food Plan defined by the U.S. Department of Agriculture in 1961: the funds needed for “Temporary or emergency use when funds are low.” The multiplier of 3 was based on the average family of two or more spending one-third of their after-tax income on food, as indicated in the 1955 Household Food Consumption Survey; the multiplication of the food budget by 3 indicated the income necessary to support that level of food consumption. While this approach did yield a poverty threshold stated in dollars, it rests on an arbitrary basis. With this threshold La Follette School Professor Rebecca Blank is chancellor of the University of Wisconsin–Madison. Prior to becoming chancellor, she spent four years in top positions of the U.S. Department of Commerce, including service as deputy secretary and as acting secretary for more than a year. Before her government service, she served as dean and professor of public policy and economics in the Gerald R. Ford School of Public Policy at the University of Michigan from 1999 to 2008. 8 / La Follette Policy Report as a starting point, an equivalence scale based on relative food expenditures among different family types was developed to produce poverty thresholds for families of different sizes and configurations. Over time, these poverty thresholds have been updated annually by changes in the consumer price index, which reflects the price of a basket of goods a typical urban consumer purchases. Aside from these annual threshold updates, there have been very few changes to the 1965 measure, which is often referred to as the “official” poverty line. To calculate the scope of poverty, “family resources” must be defined to compare them to the poverty threshold to determine if a family is or is not poor. Orshansky used a family’s pretax cash income; hence, a family whose pretax cash income fell below the poverty threshold for a family of their size and configuration would be considered poor. The overall poverty rate was calculated as the total number of people living in families whose income was below the poverty threshold, divided by the total population. Separate rates were calculated for different subpopulations (such as by race and ethnicity, age, gender, or family composition). This poverty calculation has been used since the mid-1960s and is the nation’s official poverty measure. It is an “absolute” measure so that if low-income families experience real income growth, an increasing number of them will move above this threshold and the poverty rate would automatically fall. In 1963, the poverty line was about 49 percent of median income, but by the early 2000s it had fallen to less than 30 percent of median income. The original presentation of these official poverty rates occurred in the 1964 Economic Report of the President. The results of Orshansky’s measurement efforts are shown in the solid line of Figure 1. While 22.4 percent of the population was poor in 1959, this number fell rapidly to a low of 11.1 percent in 1973. It has never been this low in any year since. After 1973, the poverty rate stagnated, rising in recessions and falling in good times. The last year for which data are available is 2013, when the official poverty rate was 14.5 percent, much higher than in the late 1960s and early 1970s. In short, official poverty rates have shown no noticeable long-term trend in the United States—and, in particular, have not declined—for 40 years despite the poverty line as a percentage of median income falling nearly 20 percentage points. www.lafollette.wisc.edu Fall 2015 Figure 1. Poverty Rates, 1969-2012 20% Supplemental Poverty Measure 15% Official Poverty Measure 10% 5% 0% 1969 197419791984198919941999200420092014 Source: Supplemental Poverty Measure data from Fox, L., Garfinkel, I., Kaushal, N., Waldfogel, J., & Wimer, C. (2015).Waging war on poverty: Poverty trends using a historical supplemental poverty measure. Journal of Policy Analysis and Management, 34; official poverty measure data from the Current Population Survey Annual Social and Economic Supplement. Figure adapted from Haveman, R., Blank, R., Moffitt, R., Smeeding, T. and Wallace, G. (2015), The war on poverty: Measurement, trends, and policy. Journal of Policy Analysis and Management, 34. Growing Criticisms of the Official Poverty Measure Some Progress Within a decade of its creation, the official poverty measure began to be criticized and suggestions for alternative measurement approaches began to be heard. All of its components were questioned—the definition of the resource-sharing unit, the use of pretax cash income, and the thresholds (which were thought to be too low). With improved data, analysts doubted that the threshold should be benchmarked to food alone. Moreover, the official poverty thresholds were criticized for not reflecting substantial differences in the costs of living across locations or increases in standards of living. The absolute nature of the U.S. poverty threshold, based on data from the 1950s and adjusted for inflation over time, means that there is no conceptual justification to the current poverty line; it is simply an arbitrary dollar amount. Using a threshold calculated in 1964 (based on 1950s data) to estimate poverty in 2014 is to use a 50-year-old categorization. Because this measure is based on cash income, it is not affected by the many in-kind antipoverty programs initiated in the United States during the past five decades. Because tax measures do not affect the definition of pretax income, substantial increases in after-tax income among low-income families due to the several expansions of the earned income tax credit over the years had no impact on the measure of poverty. In essence, the very definition created in the Johnson administration to help understand poverty has led to serious misunderstandings because of its growing inadequacy over time. Fall 2015 Over the years, a variety of formal efforts recognized some of these issues in attempts to update and refine a new and improved poverty measure. In the early 1990s, the poverty measure was formally reviewed by a panel created by the National Academy of Sciences. The panel recommended alterations to the definition of the poverty threshold and adjustment of these thresholds for cost-of-living differences across regions and rural/urban areas. Moreover, the panel recommended a resource definition that measured after-tax income (since taxes are mandatory payments) plus imputed in-kind benefits from major near-cash programs (primarily Food Stamps and housing assistance). The value of health insurance was not imputed but the panel recommended that out-of-pocket expenditures on health care be subtracted from after-tax income, since these resources are not available to be spent on food, shelter, and clothing. Work-related expenses, including child care, were also proposed to be subtracted from resources; these were treated as necessary expenditures in order to earn a living. The National Academy of Sciences recommendations led to a substantial body of follow-up research and a few cities or regions have implemented a version of the recommendations. Starting in 2011, the U.S. Census Bureau began to regularly report a supplemental poverty measure, which is loosely based on the academy recommendations; other alternative poverty measures using more expansive resource definitions, taking account of taxes and in-kind benefits, were also published. But neither the supplemental poverty measure nor alternative www.lafollette.wisc.edu La Follette Policy Report / 9 poverty rates published by the Census Bureau generated sufficient support in Congress to revise the official poverty measure. Figure 1 shows the supplemental and official poverty measures. In addition to pretax cash income, which is the basis for the official measure, the supplemental measure takes into account in-kind benefit programs and benefits conveyed through the tax system in the resource measure. The supplemental poverty measure also deducts work-related expenses and out-of-pocket health-care expenses from income. Because the supplemental measure poverty thresholds are based on expenditures on food, housing, and clothing (rather than just food) and are adjusted over time as the composition of expenditures changes, the supplemental poverty measure is a quasi-relative poverty measure. The measure accounts for differences in housing costs among areas, and it uses an improved equivalence scale to determine thresholds for different types of families. The supplemental measure indicates that poverty has declined over time, rather than being essentially flat as the official measure implies. Alternative Approaches to Poverty Measurement The supplemental poverty measure is an effort to update poverty measurement, but its approach is conceptually similar to Orshansky’s with a poverty threshold based on expenditures on necessities and a resource measure based on family resources. Other approaches have been proposed to measure poverty in fundamentally different ways. For example, many have suggested that poverty be defined as the share of the population below some point in the income distribution. Researchers proposed in 1967 and in 1990 a poverty threshold be set at 50 percent of median income. In contrast to the absolute poverty lines used in the official measure, a relative measure of poverty would remain constant even if all incomes are growing proportionally across the distribution. An alternative approach is to define a poverty threshold by estimating the cost of a comprehensive basket of necessary expenditures. Rather than using data on expenditures to determine a poverty threshold, such an approach would require an objective determination of what is “necessary” and what is a “reasonable cost” for those things deemed necessary. Efforts have been made in the United States to create such baskets. The European Union has recently funded several major research projects to create low-income expenditure baskets and compare the resulting poverty measure to other approaches. A further alternative focuses on using expenditure data rather than income data in calculating the resource side of the poverty measure. Of course, this method requires reliable measures of family expenditures (as opposed to income), which are not as frequently collected in many countries (although the United States has an annual expenditure survey). The U.S. 2014 Economic Report of the President suggests the supplemental poverty measure shows similar trends to an expenditure-based measure. A final alternative is to measure material hardship directly, rather than to assume that it is created by low income levels. Material hardship measures are only imperfectly correlated with income and clearly provide additional information about economic need. The European Union has adopted this approach, which supplements a poverty measure with multiple other measures of deprivation. The European Union requires each member state to report on 14 indicators of social exclusion, including a relative poverty measure, labor market, child well-being, and health outcomes. u Top Policy and Management Scholars The La Follette School welcomes to its faculty u Emilia Tjernström, whose dissertation at u Rourke O’Brien, who is completing a University of California, Davis, employed field Robert Wood Johnson post-doctoral experiments to examine the impact of food fellowship at Harvard University. policy innovations in developing countries. Our faculty include three fellows of the National Academy of Public Administration u Robert J. Lavigna u Donald Moynihan 10 / La Follette Policy Reportwww.lafollette.wisc.edu u David Weimer Fall 2015 Racial Variation in the Effect of Incarceration on Neighborhood Attainment By Michael Massoglia T he U.S. prison population has quadrupled since the mid1970s, leaving the United States with the highest incarceration rate in the world. This dramatic expansion reflects one of the largest policy experiments of the 20th century, and researchers and policymakers are just beginning to understand the impact this experiment has had on U.S. society. Imprisonment rates are higher for African Americans and Hispanics than for whites; indeed, the incarceration rate for blacks is more than six times larger than the rate for whites. Incarceration has become an increasingly common part of the life course, especially for black males with low levels of education. It is reasonable to assume that rising imprisonment has contributed to existing racial inequalities in U.S. society. In fact, studies indicate that imprisonment has disproportionately disadvantaged minority ex-inmates, their families, and their communities. Disproportionate incarceration plays a role in racial variation in earnings as well as certain aspects of health. Additionally, felon disenfranchisement, or the restriction of voting rights among ex-offenders, disproportionately affects African Americans, which has major implications for state and federal elections. Finally, because of large racial discrepancies in incarceration rates, black children are actually more likely to have an incarcerated mother than white children are to have an incarcerated father. Recent research suggests that minority ex-inmates also may be disadvantaged in the residential environment. Racial and ethnic minority ex-inmates live in poorer and more disadvantaged neighborhoods after prison as compared to white ex-inmates. These studies are limited, however, by their inability to account for the quality of the neighborhood in which prisoners lived prior to incarceration. In fact, the quality of the neighborhood of origin for the typical minority prisoner Michael Massoglia is a professor of sociology at the University of Wisconsin–Madison. He spoke at the La Follette School’s spring symposium “Urban Men in Poverty: Problems and Solutions” in Milwaukee. This article is adapted from an American Sociological Review piece he wrote with Glenn Firebaugh, a professor of sociology and demography at Pennsylvania State University, and Cody Warner, an assistant professor of sociology at Montana State University. Fall 2015 is worse than the neighborhood of origin for the typical white prisoner. In 1980, for example, the average minority person in U.S. urban areas lived in a lower quality neighborhood (as measured by the poverty rate) than nine out of 10 whites. Given the magnitude of the neighborhood racial divide, whites tend to have more to lose than minorities from a spell of confinement. Incarceration is much more unusual in white communities than in black communities. Because neighborhoods where incarceration is unusual are less likely to welcome their straying members, whites might be less inclined than African Americans to return to their neighborhoods of origin. Whether this neighborhood response will result in a white released prisoner moving to a poorer neighborhood is an open question. Although blacks reside in the poorest neighborhoods after prison, we do not know whether this pattern reflects an incarceration effect or existing racial residential inequalities. Hence, in this study we attempt to measure the effect of incarceration on residential quality controlling for the quality of the preimprisonment neighborhood. Does this effect of incarceration vary by race? To answer this question, we use a unique nationally representative longitudinal dataset that allows us to track individuals as they transition between prisons and communities across roughly 30 years. We first describe the types of neighborhoods where exinmates reside. We then discuss residential attainment and on the consequences of incarceration, developing testable hypotheses on incarceration’s impact on neighborhood quality. We discuss our data and methods. The key here is our use of fixed-effects models that enable us to estimate the impact of incarceration while accounting for prisoners’ neighborhoods of origin. We find that incarceration has a negative effect on neighborhood attainment only for whites and not for minority ex-inmates or even ex-inmates as a whole. After examining this effect further, we conclude by noting the importance of a deeper and more nuanced understanding of the lasting effects of incarceration on America’s growing felon class. The Neighborhood Quality of Ex-Inmates As a group, individuals who have been incarcerated live in less desirable neighborhoods than do individuals without histories of incarceration. The best evidence of this comes from the Returning Home Project, in which researchers tracked released offenders across several metropolitan areas. More than one-half www.lafollette.wisc.edu La Follette Policy Report / 11 of the released inmates followed in Chicago settled in seven of ex-inmates might be explicitly targeted and excluded from the 77 community areas (aggregates of tracts that reflect neighsome neighborhoods or communities. borhoods). High rates of poverty and disadvantage typify these For the nearly 80 percent of prisoners who are released with seven areas. supervisory parole, the close monitoring of living arrangements Evidence that post-prison neighborhood environment may create barriers to finding adequate and stable housing. affects recidivism and that one’s residence shapes one’s life sugCorrectional agencies often require pre-approval of housing gests the importance of ascertaining ex-inmates’ residential choices, and in many respects housing discrimination against destinations. Indeed, given the former inmates is now legalAs a group, individuals who have large racial disparities in conly sanctioned. For example, finement, growth in the prison individuals convicted of drug been incarcerated live in less desirable population may have important crimes can be banned from implications for racial inequali- neighborhoods than do individuals without public housing, which, ironities across a number of dimencally, is specifically intended histories of incarceration. Evidence that sions tied to neighborhood to provide assistance to those post-prison neighborhood environment context (e.g., health and labor most in need of housing. Moremarket outcomes) as an outover, as an economically maraffects recidivism and that one’s residence growth of its (presumed) effect ginalized subgroup, ex-inmates shapes one’s life suggests the importance on neighborhood attainment may also encounter commeritself. cial rental agencies that simply of ascertaining ex-inmates’ residential The observed association refuse to rent to them. Faced destinations. between incarceration and with such discrimination, neighborhood attainment does many ex-inmates may have few not necessarily reflect a causal relationship. Because ex-inmates options outside the most disadvantaged neighborhoods. are more likely to be male, young, poor, unemployed, a racial Given these considerations, our first hypothesis is: Conor ethnic minority, and have a low level of education than other trolling for neighborhood of origin and other determinants of citizens, we statistically control for these factors in measuring residential location, ex-inmates will tend to reside in more disthe causal effect of incarceration on neighborhood quality. advantaged neighborhoods following release from prison than However, controlling for these individual characteristics is the neighborhoods in which they lived prior to incarceration. insufficient to ensure that we have estimated the causal effect Because of the considerable racial disparities in both patof incarceration on neighborhood disadvantage. It is also necterns of neighborhood quality and rates of incarceration, this essary to control for the quality of the neighborhood in which general expectation is likely to be a complicated one. In parreleased prisoners lived prior to prison. ticular, we attempt to determine if the neighborhood quality We therefore account for both individual traits and neighconsequences of imprisonment will be greater for individual borhood of origin prior to prison in estimating the effect of whites or for individual racial/ethnic minorities. While prior incarceration on neighborhood quality. Our data are a combistudies find that black parolees live in neighborhoods with nation of individual data from the 1979 National Longitudinal more concentrated disadvantage and more residential instaSurvey of Youth and contextual (tract-level) data from the U.S. bility than white parolees, they reach this conclusion without Census. information on neighborhood of origin. Given extensive racial residential inequality, incarceration may do little to actually change the neighborhood trajectories of minority ex-inmates, Our Research Hypotheses while white inmates have more to lose given their advantaged Because the quality of one’s neighborhood is an established starting points. Hence, our second hypothesis is that the incarmarker of social standing, Americans are willing to pay more ceration effect on neighborhood attainment will differ for Afrito live in more desirable neighborhoods. Although incarceracan Americans, Hispanics, and whites. tion is rarely considered in studies of neighborhood quality, we expect that incarceration does affect the quality of neighborData hoods. For example, as a form of coercive mobility, incarceration, at least temporarily, forcibly removes individuals from Our analysis uses the 1979 National Longitudinal Survey of their communities. Upon release, ex-inmates might experience Youth, a data collection that began with 12,686 individuals constrained residential options stemming directly or indirectly between 14 and 22 years old. Respondents were interviewed from their incarceration. Inmates suffer from fractured social yearly from 1979 to 1994 and biennially since 1994; data colties and an increased likelihood of divorce, meaning residences lection is ongoing. With permission of the Bureau of Labor prior to prison may be neither available nor welcoming upon Statistics we gained access to restricted data that identify release. Moreover, because incarceration limits employment respondents’ geographic locations (state, county, and census opportunities and depresses wages, ex-inmates are often unable tract) at each wave of data collection. As a result, we were able to afford to live in more desirable neighborhoods. Finally, to construct individual residential histories that cover almost 12 / La Follette Policy Reportwww.lafollette.wisc.edu Fall 2015 three decades. Treating census tracts as proxies for neighborhoods, we merged the extensive individual-level 1979 National Longitudinal Survey of Youth data with characteristics of respondents’ neighborhoods (e.g., rates of neighborhood poverty and unemployment). This combination of individual and neighborhood data covering almost 30 years provides us with a unique and high quality best dataset for examining individual and neighborhood characteristics before and after prison. racial variation in residential attainment—white respondents reside in the best neighborhoods, black respondents residing in the most disadvantaged neighborhoods, and Hispanic respondents in-between. Our data further reflect racial variation in many of our control measures. Whites have higher levels of educational attainment, homeownership, and employment than do African Americans, and are less likely to live in public housing and to have incomes below the poverty line. When we observe the simple difference between neighborhood disadvantage and ex-inmate status and race/ethnicity, Measurement two findings stand out. First, we find striking racial disparities in neighborhood attainment, with blacks and Hispanics who Incarceration and Ex-Inmate Status have never served time in prison living, on average, in more In the 1979 National Longitudinal Survey of Youth, incarceradisadvantaged neighborhoods than whites who have been in tion status is measured yearly by a residential location indicator, prison. Second, all of the ex-inmate racial groups live in more which allows us to observe prison sentences across time with disadvantaged neighborhood environments than do individucertainty. als like them who have not been incarcerated. This pattern A total of 628 respondents who were in prison but subselends credence to our first hypothesis—there is a detrimental quently released were interviewed during the three decades of effect of incarceration on neighborhood quality—but it is not observation in the survey. Additional data restrictions leave us a definitive test. with 558 ex-inmates (288 blacks, 166 whites, and 104 HispanIn fact, the question of whether the incarceration experiics) whom we follow for an average of 5.5 survey waves followence—rather than individual characteristics or pre-prison ing release from prison. neighborhood conditions—drives this relationship between incarceration and neighborhood disadvantage remains open. Neighborhood Disadvantage To more reliably assess the effects of being incarcerated (being an ex-inmate) on neighborhood disadvantage, we focus on We measure neighborhood disadvantage using census tract estimated statistical results that reveal the effect of being an measures of the poverty rate, the joblessness rate (percentage ex-inmate while controlling for a host of variables and neighof working-age people unemployed or out of the labor force), borhood of origin factors that could confound this relationthe percentage of families that are female-headed, and the ship. These statistical findpercentage of households that ings provide little support for receive public assistance. Using We emphasize that there is substantial Hypothesis 1. They strongly these four measures, we formed and meaningful racial variation in suggest that, having controlled an index of neighborhood qualfor neighborhood of origin ity; higher values of this index incarceration’s effects across different and other determinants of resiindicate poorer quality neighlife domains. In some cases incarceration dential location, there is not a borhoods. negative effect of incarceration In our statistical estimates, apparently contributes to racial and ethnic on neighborhood quality—exwe also account for wide variinequalities. In other cases, such as the inmates do not reside in more ety of individual characteristics disadvantaged neighborhoods that can change over time, such results presented here, the incarceration following prison, compared as the number of moves across effect is more pronounced for whites. with the types of neighborneighborhoods that a responhoods they resided in before dent makes, educational attainprison. ment, poverty status, type of residence, family and employment To assess Hypothesis 2, we estimated the same statistical status, and age. In our statistical analysis, we examine how relationships for each of the racial groups. This analysis indineighborhood disadvantage changes over time for individuals cates significant racial variation in the effect of incarceration holding constant the basic characteristics of the neighborhood on neighborhood quality. Our statistical results show that, after where a person previously resided. accounting for neighborhood of origin, it is whites, not African Americans or Hispanics, whose neighborhood environments Results are most affected by a prison spell. This finding suggests that We have 2,629 valid observations of ex-inmates, and their racial for blacks and Hispanics (but not for whites), the association characteristics represent racial patterns in the incarcerated popbetween incarceration and neighborhood disadvantage found ulation—approximately one-half are African American, 30 perin the overall data is, in fact, attributable to the individual cent are white, and 20 percent are Hispanic. The characteristics traits or pre-prison neighborhood histories of the ex-inmates of the neighborhoods in which they live also reflect well-known themselves. It also suggests that the statistically insignificant Fall 2015 www.lafollette.wisc.edu La Follette Policy Report / 13 effect of incarceration on neighborhood disadvantage for all exinmates collectively masks the significant effect of incarceration for whites. In separate estimates, we also find that for white ex-inmates the adverse neighborhood effect of incarceration appears to increase over time. In summary, this additional analysis provides a more complete, and complex, picture of the incarceration–neighborhood disadvantage relationship than provided by previous research. After accounting for numerous characteristics and pre-prison neighborhood context, we find no effect of incarceration on neighborhood disadvantage for African Americans or Hispanics. In contrast, incarceration has a significant negative effect on neighborhood attainment for whites, and this penalty appears to intensify across time. To ensure that our results are robust, we performed additional statistical analyses. These ‘sensitivity studies’ address factors that could yield biased statistical results. For example, our results could be biased if the set of factors that we measure and statistically control for is incomplete. Across nine statistical tests, two facts never change. First, for whites the effect of incarceration is always adverse, and the coefficient is always statistically significant. Second, the effects of incarceration on African Americans and Hispanics are never statistically significant. We see no evidence to reject our main findings. As we have emphasized, our results suggest that whites have much more to lose with regard to neighborhood quality from being incarcerated than to blacks or Hispanics. This explanation is plausible because disparities in pre-prison neighborhood environments for whites, Hispanics, and African Americans are massive: while blacks and Hispanics live in neighborhoods that are greatly disadvantaged prior to incarceration relative to the mean, whites live in neighborhoods that are well above the mean level of quality. Our finding that whites have more to lose from a spell of incarceration than do African Americans raises an important question: Why is the incarceration penalty not more severe for whites than for African Americans in other domains where whites are also more advantaged, such as wages? The answer, we suspect, is that blacks and whites differ much more with regard to neighborhood environment than they do with regard to wages or employment. Indeed, in our study we show that in 2008 the difference in the average hourly wage for blacks and whites in the 1979 National Longitudinal Survey of Youth data was substantially less than the racial difference in neighborhood disadvantage. Discussion and Conclusions Given the dramatic swelling of the ex-inmate population in the United States, understanding the lasting effects of incarceration on ex-inmates, their families, and their communities is critical. While most research on the consequences of incarceration focuses on individual and family outcomes, much less is known about incarceration’s effect on residential outcomes such as neighborhood quality. By using nationally representative longitudinal data to examine change in neighborhood attainment across time, we discovered that white ex-inmates live in significantly more disadvantaged neighborhoods after a prison spell than they did before. We found no effect for neighborhood characteristics of ex-inmates as a group, or for African American or Hispanic exinmates. Additionally, and again for whites only, incarceration’s adverse effect on neighborhood attainment intensifies during the years following release from prison. What remains to be determined is whether the pre- and post-prison disparity for whites could reflect the number of arrests or criminal convictions as opposed to the effect of incarceration. While our data are relatively limited in terms of measures of arrests and criminal convictions, the weight of the evidence suggests that the pre- and post-prison difference we observed for whites reflects primarily the effect of a prison spell, not the effect of criminal offending or a criminal record. Incarceration automatically removes individuals from their neighborhoods; a criminal record does not. In other words, the causal chain appears to be: conviction g prison sentence g uprooted from current neighborhood g move to a new, more disadvantaged neighborhood upon release from prison. What if conviction does not lead to a prison spell? The chain of events would be different. Because conviction itself does not necessarily, or even likely, uproot an individual from his neighborhood, individuals who do choose to move are likely to do so voluntarily. However, a definitive answer to this question awaits further research. Our findings have a number of policy implications. To say that incarceration tends to harm whites more than African Americans with respect to neighborhood attainment is not to say that incarceration effects always tend to be greater for whites or are always inconsequential for African Americans. Rather, we emphasize that there is substantial and meaningful racial variation in incarceration’s effects across different life domains. In some cases incarceration apparently contributes to racial and ethnic inequalities. In other cases, such as the results presented here, the incarceration effect is more pronounced for whites. There is evidence that this is also the case for mortality and recidivism. Policymakers should be attentive to these differences in fashioning policies to temper the societal costs of mass incarceration. We noted earlier that the steep rise in the prison population is largely policy-driven, rather than being tied to any dramatic increase in criminal activity. As such, reductions in the use of incarceration must also be driven by policy. Clearly a balance needs to be struck between public safety and the costs of incarceration. In a time when federal and state budgets are being strained, many observers have started to question the current balance, noting that increased public funds directed to the correctional system come at the expense of funds for education, health, or any number of other public goods and services. Even if the prison boom has peaked, the consequences of that boom will be felt for decades to come, as large numbers of prisoners are reintegrated into U.S. society. Results presented in this article provide a strong reminder of the need for effective policies concerning that reintegration process. u 14 / La Follette Policy Reportwww.lafollette.wisc.edu Fall 2015 Director’s Perspective continued from page 1 prospects for employment and stable lives. Franklin, who is director of the Marquette University Law School Poll, also presented results from a public opinion poll of Wisconsin voters that captured opinions on several government policies, particularly those related to services for low-income people. Next Professor David Pate of the University of Wisconsin– Milwaukee Department of Social Work reported on his work over more than two decades examining the lives of men living in poverty. Professor Mike Massoglia of the University of Wisconsin–Madison Department of Sociology then described the devastating impact of incarceration and correction policy. He has found, for instance, that there are serious impacts on an individual level (jobs, wages, health); on a family level (family functioning and spousal relationships); on a community level (neighborhood impacts, civic participation and representation); and on a state level (creating and perpetuating historic patterns for the already disadvantaged). This issue of the Policy Report includes an article by him on racial variation and the effect of incarceration on neighborhoods. Finally, Harry Holzer, a Georgetown University professor of public policy, summarized the day by challenging the crowd of more than 200 to “find the gems” in the research and “figure out how to replicate them and take them to scale.” For example, Holzer pointed to programs focused on prevention (Big Brother/Big Sister and career academies) and re-connection (Job Corps, Jobs Plus) of disconnected youth before they become offenders. For offenders, he recommended transitional jobs and programs like the Strong Fathers’ Initiative in New York. Marquette University President Mike Lovell agreed with Holzer, saying “the only way we’re going to face and overcome the problems of urban men in poverty is by working together.” I could not agree more. And I hope you find this type of work by the La Follette School as important as I do. A podcast of the entire program is available at http://go.wisc.edu/4dvi04. I am especially pleased that this symposium builds on the decades of contributions the La Follette School has made in Milwaukee, including 25 years of capstone projects with the city of Milwaukee, a national conference on the future of big cities, evaluation of welfare/work programs such as Milwaukee’s New Hope and Wisconsin Works (W2), research on the Milwaukee Parental School Choice program, and dozens of other partnerships on topics ranging from gangs and youth violence to value-added education research. In the years to come, it is our pledge that efforts by the La Follette faculty, students, staff, alumni and friends will continue to help make a difference. On Wisconsin! The La Follette School of Public Affairs Congratulates University of Wisconsin–Madison Chancellor Rebecca Blank for winning the 2015 Daniel Patrick Moynihan Prize from the American Academy of Political and Social Science Fall 2015 www.lafollette.wisc.edu La Follette Policy Report / 15 The Effect of Child Support on the Work and Earnings of Custodial Mothers: New Evidence By Laura Cuesta and Maria Cancian T he poverty rate in 2012 for families in which a child lives with only the biological mother was more than four times (40.9 percent) that of two-parent families (8.9 percent), and almost twice the poverty rate observed among custodial-father families (22.6 percent), according to the U.S. Census Bureau. Because cash assistance for low-income families has declined dramatically in the last decade, and receipt of child support (i.e., money payments from parents who do not live with their children) is often small and irregular, the earnings of these mothers are a crucial source of income for these families. Given that child support payments also contribute to the income of these families, observers suppose the decision of whether to work likely depends on whether mothers receive child support. Indeed, they may think the receipt of child support is a substitute for earnings by the mothers, who then reduce their work and earnings. Observers tend to suppose that the more child support a custodial mother receives, the less she is likely to work for pay. Our new research published in the July 2105 Children and Youth Services Review suggests otherwise, and policymakers who focus on increasing child support payments by non-custodial parents may want to take note. Most previous research suggests that child support has a small, negative effect on the work and earnings of custodial mothers. However, this research is limited in that it fails to adequately control for difficult-to-observe characteristics of mothers that may also affect their willingness to work—things such as the mothers’ motivation, self-advocacy, and perceptions about receiving welfare benefits. As we note in our Children and Youth Services Review article, the failure to adequately account for these unobserved factors that may also influence mother’s work and earnings suggests that the estimates in the literature may not be “causal.” Another weakness in this research is that never-married mothers are excluded from these analyses. Because these women face rather different incentives to work— largely due to policies focused on them—the overall estimates may be affected by their exclusion. Our study in Children and Youth Services Review takes advantage of a statewide randomized experiment to provide more convincing estimates of the effect of child support on the employment of low-income, custodial mothers. For custodial mothers who participate in the Temporary Assistance for Needy Families (TANF) program, we examine the effect of child support on their work. Does the receipt of child support payments reduce the likelihood of working, or the hours worked, for these mothers? Laura Cuesta is assistant professor of social work at Rutgers University. She is interested in the role of public policies in the wellbeing of disadvantaged children and families. Her current research looks at the effects of child support policy; international approaches to child support; and the interaction of child support policies with welfare programs. Her dissertation examined the role of child support on Laura Cuesta the wellbeing of custodial-mother families in the United States and Colombia. She received her doctorate in social welfare from the University of Wisconsin–Madison. Maria Cancian is professor of public affairs and social work at the University of Wisconsin–Madison. Her research considers the relationship between public policies and changes in marriage, fertility, and employment, with a focus on the implications of child support policy for the well-being of divorced and never-married families, the employment and income of women who have received welfare, and the impact of married women’s growing employment and earnings on marriage Maria Cancian patterns and the inter- and intra-household distribution of income. She received her doctorate in economics from the University of Michigan. This article is a summary of a paper recently published in Children and Youth Services Review. Policy Context With the enactment of welfare reform in the mid-1990s, the United States moved from cash income support for (mainly) poor single mothers to the promotion of self-sufficiency. TANF created this focus on mothers’ employment by conditioning cash assistance on parents’ work efforts. In Wisconsin, the TANF program (called W-2) attempts to recreate some features of the job market as part of a strategy to promote work. For example, to promote employment among participants, W-2 cash assistance does not vary with family size or the mothers’ hours of work. As discussed in our Children and Youth Services Review article, the Wisconsin TANF program has other characteristics, including the approach to child support for families receiving cash assistance. Unlike the program in other states, in Wisconsin, 16 / La Follette Policy Reportwww.lafollette.wisc.edu Fall 2015 most custodial mothers in the original W-2 program were allowed to receive all child support paid on behalf of their children. Moreover, Wisconsin ignored child support income when determining TANF eligibility. This policy was consistent with the W-2 emphasis on replicating the job market. Just as workers’ wages are not affected if they receive child support, neither would child support reduce W-2 cash payments. In addition to this policy differing from that in other states, it contrasted with prior Aid to Families with Dependent Children (AFDC) policy. AFDC allowed for a maximum $50 child support to be passed through; only $50 was also disregarded in determining AFDC cash benefits. Most other states maintained this $50 passthrough and disregard, or eliminated it completely. Because Wisconsin’s approach was more generous than the norm, the federal government required the state to apply for a waiver. As a condition for granting the waiver, the federal government required an experimental evaluation of the policy. Wisconsin’s approach, and the required evaluation, provided the opportunity to evaluate the effects of child support on a host of individual and family outcomes. In the fall of 1997, the experiment was inaugurated; both recipients of AFDC transitioning to TANF and new applicants to TANF were randomly assigned to one of two pass-through eligibility statuses. Participants in the experimental group received the full amount of child support paid by the noncustodial parent, while participants in the control group received a partial pass-through of the first $50 per month, or 41 percent of the amount paid, whichever was larger. Our Children and Youth Services Review article considers mothers’ labor supply during the initial experiment beginning in 1997. Later, in July 2002, all custodial parents began to receive the full pass-through of child support. In 2003, with the end of the federal waiver that required the experiment, Wisconsin stopped allowing custodial parents on TANF to receive all child support paid on behalf of their children. Prior Research Given the importance of child support income and mothers’ earnings for the economic well-being of low-income, custodialmother families, it is surprising that the literature that examines policies that simultaneously encourages child support collections and mothers’ employment is so small. The few earlier studies have considered the effect of child support income on the work effort of divorced and separated mothers who were both recipients of AFDC benefits and nonrecipients. Moreover, as noted in our Children and Youth Services Review article, this research has come to somewhat different conclusions. While research by John Graham and Andrea Beller published in the late 1980s suggests that child support is negatively associated with the number of hours worked, the negative effect was only one-third to one-half the effect of other sources of income other than earnings. This smaller effect is likely due to the fact that child support is a risky income source, leading mothers to work and earn to offset this risk. Research by Wei-Yin Hu published in 1999 shows a small, negative effect of child support on the labor supply of mothers who are not on AFDC (and, hence, Fall 2015 likely to be better off than a welfare sample) and a positive effect among mothers receiving cash welfare. Hu found that child support increases the number of hours mothers on AFDC worked. Methods In our Children and Youth Services Review study, we use survey and administrative data from the Wisconsin Child Support Demonstration Evaluation, which uses an experimental design to evaluate the effects of child support on several outcomes. The evaluation study collected three waves of information about the custodial mother, one child (referred to as the “focal” child), and that child’s noncustodial father. Because custodial mothers may make decisions about work and earnings only after they observe a regular pattern of child support payments or an increased amount of child support income, the effects of the experiment may take some time to be observed. Hence, we base our study on employment and earnings two years after the 1997 starting date (i.e., between the last month of 1999 and the first half of 2000). The final sample includes 2,085 mothers who were interviewed in 1997 and two years later. Our data provided comprehensive information on employment history, child support, demographics, and socioeconomic characteristics of the mother and her family. When randomized experiments are perfectly implemented, the simple comparison of mean values between experimental and control groups provides the unbiased causal effect. However, no randomized experiment is free from some randomly occurring differences in initial characteristics, and this experiment is no different. To accommodate this problem, we rely on statistically adjusted means rather than simple means. We proceed from more simple and direct estimates of the effect of child support receipt on mothers’ employment and hours worked to estimates that attempt to adjust for the fact that child support receipt and employment are interdependent. The simple and direct estimates provide a context for the more advanced analyses that exploit the experimental design in the Children and Youth Services Review study. All models include other variables than simply the receipt of child support payments. These variables include the mother’s age, race, education, her AFDC and employment history, child support receipt history, number of children, and residency. In addition, the focal child’s age, the noncustodial father’s average annual earnings, the number of legal fathers associated with the mother are included. Results The straightforward estimates directly compare mothers who receive child support and those who do not. They suggest that the mothers who receive child support are more likely to work (62.5 percent) than those who do not (53.2 percent), a difference of 9.3 percentage points. These estimates also suggest that, on average, custodial mothers receiving child support work three more hours per week than nonrecipients. More complex estimates that include control variables show that mothers who received any child support in 1999 are 6.1 percentage points more likely to work than those who did www.lafollette.wisc.edu La Follette Policy Report / 17 not—the positive “effect” of child support on a mother’s participation in the labor market remains even after controlling for observed characteristics. The estimates also suggest that hours worked are positively related to the receipt of child support. On average, mothers who received child support are estimated to work between 1.7 to 2.4 hours more per week than those who did not receive child support. Findings from the straightforward estimates suggest that child support has a positive and statistically significant relationship with custodial mothers’ labor supply. The findings in our Children and Youth Services Review study are consistent with those reported in Hu’s 1999 nonexperimental study. However, these estimates may be biased since they are unable to capture differences in unobserved characteristics that may influence child support receipt and participation in the labor market. In our experimental estimates, we take advantage of the random assignment of mothers to experimental (full pass-through of child support) and control groups to estimate the effect of child support on custodial mothers’ labor supply. While those in the experimental group have a higher probability of working, this estimate is not statistically significant. This positive estimate strongly suggests that child support receipt does not reduce the likelihood of working for pay among mothers participating in W-2. The estimated effect of being in the experimental group on hours worked is also positive but again is not statistically significant; this result also suggests that the receipt of child support does not reduce the number of hours worked for these mothers. As we report in Children and Youth Services Review, one explanation of this result is that mothers do not take into account child support payments in making their decisions regarding employment. The risky nature of child support payments suggests that this may not be an irrational decision. To assess this possibility we estimated the effect of child support payment receipt on mother’s employment for those mothers who are likely to have regular and reliable payments (e.g., at least $1,000 of child support was paid in an earlier year). The results did not change substantially, suggesting that this explanation may not hold. Discussion and Conclusions The overarching question of this study was whether child support receipt reduces the work effort of custodial mothers participating in TANF. Results from our direct estimates are consistent with previous analyses for custodial mothers receiving cash welfare; when unobserved differences between mothers who do and do not receive child support are not taken into account, additional support is positively associated with hours worked as in Hu’s 1999 study. However, once the variation in child support received associated with random assignment to alternative pass-through and disregard policies is used, there is no discernible effect of child support receipt on the likelihood of working for pay or hours worked. The difference between our results in Children and Youth Services Review and those of prior analyses may be due to the confounding of child support effects with other unobserved characteristics that affect mothers work. Alternatively, findings from prior research may reflect the behavioral response of custodial mothers who are relatively better off than a welfare sample. The vast majority of the sample in our study comprises mothers who have never been married, while other studies focused on divorced and separated mothers and explicitly excluded never-married mothers. Moreover, we examine this issue after work-focused welfare reform, which may have changed work incentives among low-income mothers. Our findings should be interpreted cautiously, as we note in Children and Youth Services Review. First, our results cannot be generalized to all custodial-mother families as child support payments are particularly irregular and small among disadvantaged families like those participating in TANF. As a result, the custodial mother may be more likely to make decisions about labor supply without considering potential monetary contributions from the noncustodial father; small and irregular amounts of child support may actually underscore the importance of working for pay among these women. Second, there is one characteristic of the Wisconsin experiment that may make it a weaker test. Although custodial mothers in the experimental group received on average higher amounts than those in the control group, the difference was small. Hence, while we were able to address selection into child support receipt by taking advantage of the random assignment into experimental and control groups, the actual treatment (child support income) was relatively modest. It is noteworthy, however, that this limited treatment is arguably a strength for a policy-relevant test: low-income women are likely to receive low and irregular payments. Third, by the time the experiment was conducted, Wisconsin had built a relatively strong institutional framework supporting work-focused welfare interventions; this environment may have created unique conditions that strengthen incentives to participate in the labor market and helped to avoid the negative effects of child support income on mothers’ employment. It is important to note, though, that most states have adapted to changes introduced by welfare reform; the Wisconsin experience may be relevant for other states. Finally, there is one consideration that we have not been able to effectively deal with. It may be the case that the payment of child support by the noncustodial father may be affected by the work and earnings of the mother. If they are, one could expect, for instance, that the more the mother works and earns, the less regular would be the father’s payment. This possible effect would tend to bias our results downward; the true estimate of the effect of child support on mothers’ work would be more positive than we report and potentially significant. In spite of these caveats, our results have implications for social welfare policies in the United States. Some states have implemented programs aimed at improving the employability of noncustodial fathers, in hope that these interventions will ultimately result in more steady and increased child support payments. As we note in Children and Youth Services Review, our findings suggest that this effect would not likely be offset by changes in mothers’ employment, even in the context of a full child support pass-through. Policies encouraging child support collections and mothers’ employment are not necessarily incompatible; efforts to increase the labor supply of both parents are likely to be worthy. u 18 / La Follette Policy Reportwww.lafollette.wisc.edu Fall 2015 Award-winning faculty. Cutting edge research. The La Follette School congratulates Donald Moynihan for his election as president of the Public Management Research Association. He is the winner of the Association for Public Policy Analysis and Management’s 2014 Kershaw award, which honors a scholar younger than 40 who has made a distinguished contribution to public policy analysis and management. Public Administration Review included two of his articles among the 75 most influential articles published in its 75-year history. Quality Research in the Best Journals Funded Research Top rankings in public affairs publications: University ranked 2nd for articles faculty published in the four “top” public affairs journals for 2009 to 2013 and 4th for public administration research. Publications in top field and David Weimer disciplinary journals, including recent publications in Health Affairs and the Journal of Health Politics (David Weimer), American Political Science Review (Susan Webb Yackee), American Journal of Political Science (Donald Moynihan), Journal of Health Economics, American Sociological Review, and Proceedings of the National Academy of Science (Jason Fletcher), Public Administrative Review and Journal of Public Administration Research and Theory (Pamela Herd), JAMA Pediatrics (Barbara Wolfe). Fall 2015 The La Follette School’s interdisciplinary researchers develop innovative projects with support from federal and private funding sources: nearly $35 million in National Institutes of Health funding (Pamela Herd); $1.2 million in funding from the U.S. Susan Yackee Treasury (J. Michael Collins); $500,000 from the Burroughs Wellcome Fund (Susan Webb Yackee); $460,000 from the National Institute of Child Health and Human Development, $350,000 from the William T. Grant Foundation, $92,000 from Russell Sage, (Jason Fletcher); $397,000 from the Smith Richardson Foundation, and $100,000 from the Annie E. Casey Foundation (Barbara Wolfe). www.lafollette.wisc.edu La Follette Policy Report / 19 La Follette Policy Report Robert M. La Follette School of Public Affairs University of Wisconsin–Madison 1225 Observatory Drive Madison WI 53706 Nonprofit Org. U.S. Postage Paid Madison, WI Permit No. 658