Show Summary Details

Page of

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, ECONOMICS AND FINANCE (oxfordre.com/economics). (c) Oxford University Press USA, 2019. All Rights Reserved. Personal use only; commercial use is strictly prohibited (for details see Privacy Policy and Legal Notice).

date: 17 July 2019

Poverty and Social Policy in the United States

Summary and Keywords

The interaction between poverty and social policy is an issue of longstanding interest in academic and policy circles. There are active debates on how to measure poverty, including where to draw the threshold determining whether a family is deemed to be living in poverty and how to measure resources available. These decisions have profound impacts on our understanding of the anti-poverty effectiveness of social welfare programs. In the context of the United States, focusing solely on cash income transfers shows little progress against poverty over the past 50 years, but substantive gains are obtained if the resource concept is expanded to include in-kind transfers and refundable tax credits. Beyond poverty, the research literature has examined the effects of social welfare policy on a host of outcomes such as labor supply, consumption, health, wealth, fertility, and marriage. Most of this work finds the disincentive effects of welfare programs on work, saving, and family structure to be small, but the income and consumption smoothing benefits to be sizable, and some recent work has found positive long-term effects of transfer programs on the health and education of children. More research is needed, however, on how to measure poverty, especially in the face of deteriorating quality of household surveys, on the long-term consequences of transfer programs, and on alternative designs of the welfare state.

Keywords: poverty measurement, inequality, welfare, tax credits, social insurance, means-tested transfers

Introduction

The interaction between poverty and social policy is an issue of longstanding interest to economists and policymakers, and often contentious in public and academic debates. Indeed, a frequent claim by conservative commentators in the United States is that America’s “War on Poverty” failed because poverty rates are not much different than those of 50 years ago at the dawn of the federal effort to fight poverty. Such claims raise important questions for economists on how poverty is measured, whether and how assistance programs affect household decisions and thus have a negative feedback on poverty, and whether poverty rates per se are too narrow a metric to gauge the effectiveness of the safety net.

The aim of this article is to summarize trends in and research on interactions between poverty and the social safety net, and to highlight areas where more research is needed to better understand linkages between poverty and social policy. The context for much of the institutional setting is the United States. The emphasis on the United States is based in part because America was the first country to systematically measure poverty on an annual basis, and with the passage of President Johnson’s Great Society anti-poverty programs in the mid-1960s, a proliferation of large-scale demonstration projects, social surveys, and research followed. In addition, the social safety net in the United States is massive. With spending in recent years exceeding $2 trillion, and adjusted for inflation, it has grown by more than a factor of four in per-capita terms since 1970. However, increasingly those funds have tilted away from populations historically served by anti-poverty programs such as single-mother families and toward the elderly and disabled (Moffitt, 2015; Ziliak, 2015a). This suggests that the anti-poverty effectiveness of the U.S. safety net may have deteriorated, or at least perhaps the composition of those lifted out of poverty may have shifted over time.

The section “Measuring Poverty” provides an overview of poverty measurement in America. Although some of the institutional details are unique to the United States, many of the issues raised here are ubiquitous across developed and developing countries.1 For example, a common challenge across countries is with poverty measurement—where and how to draw the line and how to measure resources. The United States adopted an absolute poverty measure in the 1960s, and while most of the other nations in the Organization for Economic Co-operation and Development (OECD) use a relative measure, developing nations also anchor their population poverty status with an absolute measure ($1.25 per person per day). There are also differences across nations on whether to measure resources as income or consumption, whether resources should be aggregated to the family or household level, whether income should be before- or after-tax measures, and whether and how to value in-kind transfers. The United States adopted a before-tax measure of income exclusive of in-kind transfers as the measure of resources. To understand the implications of this adoption, the “Social Safety Net and Poverty in America” section provides an overview of the major programs in the U.S. safety net, highlighting populations served, key program parameters, financing, and trends in spending over time. Evidence is presented on how pre-tax and transfer poverty trends are affected first by cash welfare and social insurance, and then with the addition of selected in-kind transfers and tax payments and credits to examine a more comprehensive view of the safety net’s effect on poverty.

This sets up in the penultimate section, “Research Needs on Poverty and Social Policy,” a discussion of future research needs on poverty measurement, and how the design and financing of the safety net might improve both program integrity and anti-poverty effectiveness. Active lines of research remain on the relative merits and pitfalls of absolute versus relative poverty measures, on whether poverty should be measured on a before-tax or after-tax basis, on whether consumption is a better metric than income for measuring well-being, and whether single indices (income or consumption) should be replaced (or at least supplanted with) multidimensional measures (Alkire, Foster, Seth, Santos, Roche, & Ballon, 2015; Citro & Michael, 1995; Meyer & Sullivan, 2012; Slesnick, 2001; Ziliak, 2006). Importantly, most nations measure poverty from household surveys, but these frequently are plagued by nonresponse and misreporting that can substantively affect the portrait of poverty, and recent advances and areas of additional research need such as incorporating administrative data systems with household surveys are explored (Bee, Gathright, & Meyer, 2018; Hokayem, Bollinger, & Ziliak, 2015). Beyond measurement, the section includes broader questions of transfer program design. For example, in the 1990s many OECD nations reformed their welfare systems for the non-disabled and non-elderly toward a more work-based safety net. A movement is afoot in the United States to expand that work focus to other programs such as food, housing, and health insurance, and yet research is lacking to guide policy on the efficacy of such a reform. Moreover, the U.S. reforms to cash welfare in the 1990s aimed toward more decentralized control of program design, and proposals to expand that to many others in the safety net are under consideration. Finally, many countries are experimenting with wholesale redesign of assistance in the form of universal basic income, and the article concludes with a discussion of the potential of basic income to affect family behavior and outcomes.

Measuring Poverty

Measuring the poverty status of individuals and families has been at the fore of research among poverty scholars for the past fifty years (Atkinson, 1987; Orshansky, 1963; Sen, 1976; Watts, 1968). Many of the ideas from this literature are captured by the index proposed by Foster, Greer, and Thorbecke (1984):

Pα=1nq=1Q(zyqz)α,
(1)

where n is the size of the population, Q is the number of poor persons (or families or households), z is the poverty threshold, and y is resources at the disposal of the unit of analysis. The parameter α,0α indicates the degree of aversion to poverty such that as a increases, there is increasing weight given to the poorest households. The most popular variants are with α=0,1,or2. When α=0, the index yields the commonly known headcount ratio or poverty rate, P0=Qn, which represents the percentage of the population that is poor; when α=1, it is known as the (normalized) income-gap ratio; and when α=2, the index yields the (normalized) squared poverty gap. By far the most commonly employed measure across countries is α=0 because the poverty rate is simple and transparent. However, it is a fairly blunt metric that does not account for the intensity of poverty, and thus fails some desirable properties of poverty measures.2 An attractive alternative when α=1 is to drop the normalization by z and to sum up the resource shortfall for the population for which yz, which yields the poverty gap, or the total resource shortfall among the poor population. For example, if income is the metric for poverty measurement, then this measure tells us how much income must be transferred to lift all poor persons out of poverty.

The poverty index in equation (1) requires the researcher (and policymaker) to not only make decisions on the value of a, but also on how and where to draw the threshold z and how to measure resources y. The threshold could be based on an absolute poverty scale such as a $1.25 per person per day as in developing countries, or based on a relative scale such as a percentage of median income as in some OECD nations. The absolute scale is based on the premise that poverty status should be judged against a fixed, objectively set minimum standard of living. The relative scale, however, treats the minimum standard of living as a function of the current socially accepted norms of consumption needs. There are advantages and disadvantages to each. Absolute scales are readily amenable to comparisons over time to assess how a society is faring, while the relative scale is less straightforward to benchmark progress against poverty because the threshold changes with the economy. On the down side, there is an inherent element of subjectivity involved in determining minimally adequate needs for use in absolute scales, and they may become dated over time compared to relative poverty lines.3

A case in point is the U.S. poverty measure. The United States adopted an absolute measure of poverty in the late 1960s, based on the seminal research of Orshansky (1963). Using data from the 1955 Household Food Consumption Survey, Orshansky found that the average family of three or more persons spent about one-third of their after-tax money income on food spending. This implies that after establishing the appropriate food budget, one could use a multiplier of 3 to establish an income cutoff for minimally adequate needs. Since the 1930s the U.S. Department of Agriculture has constructed four basic food plans of increasing cost, and the food plan adopted for the poverty measure was the least costly of the four. To move from the benchmark threshold of a two-adult, two-child family, Orshansky had to make assumptions on the food needs of different family members using equivalence scales, and this resulted in 62 separate food plans based on family structure, age, gender of the head, and whether the family resided on a farm. Because it was assumed that farm families would grow and produce some of their own food, the poverty line was lower than for nonfarm families. In 1980 the farm/nonfarm distinction, along with gender, were dropped. The poverty thresholds have been increased by the Consumer Price Index each year since 1969, holding constant (for inflation) the standard of living of the poor. In FY2017 the poverty line for a four-person family was $25,283.

The resources are typically measured in terms of income, but some argue that income at a point in time does not capture the permanent economic status of the individual compared to consumption, or wealth, and thus mischaracterizes poverty. Even if income is selected as the resource base, there is debate over whether income should be measured annually or at higher frequency, whether income should be before or after taxes and transfers, and among the latter, whether and how in-kind transfers should be counted. Moreover, it is not obvious ex ante whose income should be counted, that is, whether the measure should be at the individual level, family level, or household level.

Again the U.S. case is instructive. Economic resources for the purposes of official poverty measurement in the United States only entail highly liquid forms of money income. This includes labor market earnings, private pensions, rent, interest, dividend income, and cash transfers from related and unrelated individuals, as well as governmental and non-governmental organizations. This information is collected as part of a supplement to the monthly Current Population Survey (CPS) called the Annual Social and Economic Study (ASEC). The ASEC is fielded in March and the income information refers to the previous calendar year. Importantly, the resource measure is before tax payments and credits, and does not include in-kind transfers or capital gains and losses. The family is the basic unit of analysis for poverty measurement, where family means two or more persons residing together and related by marriage, birth, or adoption. The income of all family members is summed to yield total family income for the year, and members of related subfamilies are assigned the family income of the primary family unit. All members of the same family share the same poverty status.

The Social Safety Net and Poverty in America

To fix ideas on how treatment of taxes and transfers can have a fundamental impact on measures of poverty, and in order to set up the discussion on future research needs, this section provides a brief overview of the U.S. social safety net, and then demonstrates how the safety net has affected the level and trend in poverty over the past 35 years. Safety net programs in the United States are typically grouped into the two broad categories of social insurance and means-tested transfers. Social insurance programs are tied to employment, military service, or old age, while means-tested transfers are conditioned on low income, and often low assets, but not necessarily tied to employment or age.

Social Insurance

The major social insurance programs are Social Security Retirement and Survivors Benefits, Medicare, Disability Insurance (DI), Workers Compensation, Unemployment Insurance (UI), and Veterans Benefits. Social Security retirement is targeted to persons age 62 and older who have accumulated at least 40 quarters of covered employment in their careers, and monthly cash benefits are paid out as a progressive function of pre-retirement earnings, that is, low-wage workers receive a higher proportion of pre-retirement earnings paid out as benefits during retirement. Normal retirement age is 65 for persons born before 1938, but it has gradually increased to age 67 for those born after 1959. Eligible workers can claim benefits early, starting at age 62, but have benefits reduced by 25%. In addition, for persons who both work and collect benefits prior to their normal retirement age, the social security benefit is reduced 50% for each dollar earned above a threshold. Funding for the program comes from a payroll tax of 12.4% up to a cap, which is shared equally by the employer and employee.

Medicare is an in-kind transfer that provides health insurance for persons aged 65 and older, as well as some disabled persons under age 65, regardless of previous work history. The program covers both in-patient and out-patient care, as well as limited stays in assisted care facilities and coverage for prescription drugs. Basic coverage (part A) is available to all seniors, but most purchase expanded coverage (part B) via a monthly premium, the size of which increases in income. Like Social Security, funding comes from a payroll tax, but the rate is a lower 2.9% and it applies to all labor-market earnings.

Disability insurance was added to the Social Security Program in 1956, but eligibility is restricted to those workers under normal retirement age and who have worked in at least five of the last ten years and can no longer hold gainful employment owing to medically-certified disability. The monthly cash benefit amount is proportional to average lifetime earnings prior to disability, and it is financed out of the same Social Security payroll tax revenues that cover retirement benefits. Another disability insurance program is Workers Compensation. To qualify, a person must have a temporary or permanent work-related injury or illness that precludes working at the pre-injury job or one similar to it. Most of the benefits paid out are to cover medical expenses, though roughly one-fourth are paid out for lost wages. Workers comp programs vary from state to state, and benefits are financed by what is known as an experienced-rated tax on employers, that is, the tax rate is higher for those firms with greater workplace injuries and benefits claims. Finally, the Veteran’s Benefits program provides disability benefits to those armed forces members injured during service, and it also provides medical benefits, cash, and educational assistance to qualifying veterans.

Unemployment insurance is one of the oldest social insurance programs in the United States, and became a federal fixture in the mid-1930s, but major responsibility for program rules and administration largely rests at the state level. To be eligible for UI, typically the worker must have worked in covered employment in the first four out of the last five calendar quarters, must not have voluntarily left their job, must be able to work, and must be actively seeking work. Conditional on passing the work test, benefit amounts vary widely across states, though typically it is a function of past wages in the base period used for eligibility, subject to a cap. Normal UI receipt lasts up to 26 weeks, but the Extended Benefits Program that is triggered by Congressional action in periods of high unemployment allows for extensions up to 13 weeks, and the latter can be renewed such as in the Great Recession of 2007–2009 when UI eligibility lasted up to 99 weeks. Like workers comp, the UI program is funded by an experienced-rated tax on employers.

Means-Tested Transfers

The key means-tested transfer programs are Medicaid, Supplemental Security Income (SSI), Temporary Assistance for Needy Families (TANF), housing assistance, and the Supplemental Nutrition Assistance Program (SNAP). The other key means-tested programs that are directly tied to employment are the Earned Income Tax Credit (EITC) and the Additional Child Tax Credit (ACTC).

The Medicaid program is an in-kind health insurance program that was established in 1965 alongside Medicare, but the program is targeted to low-income and low liquid asset individuals and families. Many parameters relating to eligibility and benefit coverage are set at the state level. The majority of recipients in traditional Medicaid are single mothers and their dependent children, though the bulk of outlays are spent on poor seniors requiring hospital and nursing home care. Funding for the program is from a federal and state matching grant program, with the state’s share declining in state per capita personal income.

The SSI program, which was added to the Social Security Program in 1972, provides cash assistance to the needy aged, the blind, and the disabled, with the latter group making up the large majority of recipients and expenditures. The key distinction between SSI and DI is that SSI has no work history test, and thus low-income children are eligible, as are poor seniors who failed to accumulate sufficient work history prior to retirement to qualify for Social Security. The SSI program has substantial federal oversight, with grant and eligibility criteria set at the federal level, along with statutory benefit reduction rates on earned and unearned incomes. Most of the states supplement the federal grant for individuals living independently.

The traditional “welfare” program in the United States is TANF. It was established in the 1930s as Aid to Dependent Children, and then was first renamed Aid to Families with Dependent Children (AFDC) in the early 1960s when eligibility of adult caretakers was added to the program, and finally renamed TANF with the 1996 welfare reform that overhauled the program from a federal matching grant (using the same match rate as in Medicaid) to a fixed block grant to states. Most features of program design were devolved to the states, including benefit levels, eligibility, and the mix of cash versus in-kind assistance. Previously the program provided mostly monthly cash benefits, but under TANF most states provide the bulk of assistance in the form of in-kind benefits. While the federal government provides funds for TANF via the block grant, states are obligated to support the program out of their own funds as well.

Virtually all those who participated in the AFDC program also received assistance from the Food Stamp Program, now known as SNAP. SNAP provides in-kind food assistance to low-income and low-asset persons without regard to age and family structure, and thus the target population is broader than either Medicaid or TANF. The benefit is delivered monthly via a debit card, and it is redeemable for food from USDA certified outlets for preparation and consumption in the home. Benefits and primary income and asset limits are federally determined, though states are responsible for part of the cost of program administration and since the 1996 welfare reform states have added discretion over program eligibility.

Housing assistance in the United States is an in-kind program begun on a large scale during the 1930s Great Depression, mostly in the form of government-built housing units. Today, most of the benefits are provided in the form of (Section 8) vouchers redeemable in the private rental housing market. A feature that makes housing assistance unique in the safety net is that responsibility is devolved to local housing authorities, which number over 3,300 nationally. Individuals are expected to cover the first 30% of monthly rent, and then the voucher covers the remainder subject to a cap. Eligibility varies across housing authorities, but the basic income test requires family income to be less than some percentage of county median income, 50% being common. However, demand for vouchers far exceeds supply, and because the program is not an entitlement, waiting lists are oftentimes years long, or closed altogether.

The newest additions to the safety net are refundable tax credits. The first of these is the EITC, which was established in 1975 and is available to low-income working families and individuals. The credit first phases in as earnings increase until a maximum is reached, then the credit is held constant over a range, and finally the credit is tapered away as earnings increase beyond the maximum. The generosity of the maximum credit increases in the number of qualifying children up to three children. Funding for the credit is out of federal tax revenues, though about one-half of the states have an additional separate state EITC funded out of state revenues. The other refundable tax credit is the ACTC, which was established in 1997. Families with annual earnings of at least $3,000 are eligible for a $1,000 tax credit for each dependent child under age 17, with eligibility phasing out with income above a threshold. For those with very low incomes, or a large number of dependents, if the tax credit results in negative tax liability then they are eligible to have the balanced refunded.

Trends in Safety Net Spending

Table 1 presents spending on the largest social insurance and means-tested transfer programs from 1980 to 2010, expressed in real 2010 dollars using the Personal Consumption Expenditure Deflator. There has been a dramatic expansion of the social safety net in the United States over the past 30 years. Table 1 shows that from 1980 to 2010, inflation-adjusted spending on the EITC increased nearly 1,400%, Medicare and Medicaid increased 500% or more, spending on DI, UI, and SNAP have each increased over 200%, and there was a more than doubling of spending on Social Security, SSI, and housing. The only program with stable spending is TANF.

Table 1. Trends in real ($2010) spending on selected social insurance and means-tested transfer programs.

1980

1990

2000

2010

Social Security

243.5

336.3

429.6

522.9

Medicare

85.3

167.4

270.1

508.9

Disability Insurance

35.9

37.4

66.9

124.2

Workers Compensation

31.5

57.6

58.1

58.2

Unemployment Insurance

37.3

27.6

25.5

138.6

Veterans Benefits

34.1

26.7

30.4

51.4

Medicaid

54.0

97.9

205.1

338.8

Supplemental Security Income

17.8

24.3

37.4

48.2

Temporary Assistance for Needy Families

31.0

33.3

34.5

35.8

Supplemental Nutrition Assistance Program

21.3

23.2

20.8

68.3

Housing Assistance

21.1

27.3

39.8

42.2

Earned Income Tax Credit

4.4

11.3

39.3

60.9

Note: Expenditures are in 2010 dollars based on the personal consumption expenditure deflator.

Sources: 2013 Annual Statistical Supplement to the Social Security Bulletin (OASI and DI from Table 7.A.4; Medicare is the sum of Table 8.A.1 & 8.A.2; Medicaid from Table 8.E2 (1980 value from the 2000 supplement and 1970 value from https://www.gpo.gov/fdsys/pkg/GPO-CPRT-104WPRT23609/pdf/GPO-CPRT-104WPRT23609-2-16.pdf);

SSI from Table 7.A.4; Workers Compensation from Table 9.B1; UI data includes extended benefits and was obtained from https://workforcesecurity.doleta.gov/unemploy/Chartbook/b1.asp; Veterans Benefits for 1980 from Table 518 at https://www.census.gov/prod/2002pubs/01statab/socinsur.pdf; Veterans Benefits for 1990, 2000, 2009 from Table 540 at https://www.census.gov/compendia/statab/2012/tables/12s0540.pdf; AFDC for 1980 from Table 8-22 of 1996 Green Book at http://www.gpo.gov/fdsys/pkg/GPO-CPRT-104WPRT23609/pdf/GPO-CPRT-104WPRT23609-2-8.pdf; AFDC for 1990 and TANF for 2000 and 2010 from Table 7-2 of 2012 Green Book at https://greenbook-waysandmeans.house.gov/; Food Stamps/SNAP from http://www.fns.usda.gov/pd/SNAPsummary.htm; Housing Assistance for 1980, 1990, 2000 from Table 15.24 from https://www.govinfo.gov/content/pkg/GPO-CPRT-108WPRT108-6/pdf/GPO-CPRT-108WPRT108-6-2-15.pdf; Housing Assistance for 2011 from Table A.1 of https://greenbook-waysandmeans.house.gov/sites/greenbook.waysandmeans.house.gov/files/2012/documents/RL41823_gb.pdf; EITC from http://www.taxpolicycenter.org/taxfacts/displayafact.cfm?DocID=37&Topic2id=30&Topic3id=39

The reasons for the growth in safety net spending vary by program, but generally involve some combination of changing demographics, business cycles, and direct policy reforms (Moffitt, 2015; Ziliak, 2015a). For example, as countercyclical programs, UI and SNAP surge during economic downturns as workers lose jobs and need assistance, and the Great Recession of 2007–2009 accounts for the large outlays in UI and most (but not all) of the SNAP growth (Ziliak, 2015b). The aging of the U.S. population is the primary fuel driving growth in Social Security, Medicare, and Medicaid, while the secular decline in employment and increases in the eligible population account for the growth of DI (Autor & Duggan, 2006). And other programs, namely, the EITC, SSI, and TANF, underwent significant policy reforms. After its introduction in 1975, the EITC was greatly expanded with the tax legislation of 1986, 1990, 1993, and 2009. SSI experienced growth in the wake of welfare reform, as well as after the Supreme Court issued its 1991 Zebley Decision that greatly expanded child access to SSI (Kubik, 1999). TANF replaced the AFDC program with the 1996 welfare reform, and while the federal block grant has been fixed in nominal terms at $16.5 billion per year, real spending has remained stable because states carried forward savings from declines in caseloads, and in some cases also provided additional spending out of their own funds.

Trends in Poverty Rates

To gauge how poverty rates are affected by resources, Figure 1 shows trends in U.S. poverty rates from 1979 to 2013 (P0 from equation (1)).4 The series’ in the figure use the official U.S. poverty threshold for each year, but three income concepts for resources: (1) market income is the sum of private income sources (earnings, rent/interest/dividends, pensions, private transfers); (2) Census income is market income plus the addition of cash transfers included in the official poverty rate produced by the Census Bureau; and (3) net income is Census income plus in-kind food assistance from SNAP and tax credits from the EITC and ACTC, and less tax payments for federal, state, and payroll liabilities. Among the cash transfers that Census includes in the poverty measure are social insurance from Social Security retirement, survivors, and disability, UI, workers comp, and vet benefits, along with means-tested benefits from TANF and SSI. This means the programs that grew the most since 1980—EITC, Medicare, Medicaid, SNAP, housing—are not included in the official measure. Information on health and housing benefits are not included in the CPS ASEC, and thus are not included in the net income poverty measure. It is not obvious how to count benefits from in-kind programs like health insurance (Smeeding, 1982), and thus their omission is less controversial compared to programs like SNAP and the EITC. The unit of analysis is the family, and thus the Census income series here replicates the official poverty rate.

Poverty and Social Policy in the United StatesClick to view larger

Figure 1. Trends in person poverty rates, 1979–2013.

Over the past 35 years the average market-income poverty rate was 22%, which is 8.5 percentage points greater than the average Census poverty rate of 13.5%, and 9.1 percentage points greater than the average net income poverty rate of 12.9%. The comparison between market income and Census income highlights the strong anti-poverty effectiveness of such transfers as Social Security, DI, and SSI. It is important to point out, however, that the wedge between the market poverty rate and the official rate may also be due in part to the negative behavioral response of labor supply and saving induced by the fact that the transfers either explicitly tax wage and capital income, or implicitly tax private income, as highlighted in much of the early economics literature (Danziger, Haveman, & Plotnick, 1981; Moffitt, 1992). These negative behavioral effects, though, were generally small in magnitude. Perhaps surprising, Figure 1 shows that the net income poverty rate actually exceeds the Census poverty rate for most of the 1980s. This occurs because the negative tax liability outweighed the positive contribution of food stamps and the EITC in that period. However, after the Tax Reform Act of 1986, the expansion of the EITC in the early 1990s, the introduction of the ACTC, and growth in SNAP participation in the 2000s, that after 1994 the net income poverty rate increasingly pulled away from the Census rate. This is especially true after the Great Recession in 2008-2009.

Regardless of income measure there is a clear counter-cyclical pattern to poverty rates, rising during the recessions of 1981, 1990, 2001, and 2008, and declining during the long expansions in between. There has been considerable research on whether and to what extent the macroeconomy affects the head count rate, with all of it concluding that even after controlling for other confounding factors, poverty rates decline with both declines in the unemployment rate and increases in the employment growth rate (Anderson, 1964; Blank & Card, 1993; Gottschalk & Danziger, 1985; Gundersen & Ziliak, 2004). What also appears clear from Figure 1 is that the safety net reduces the amplitude of poverty over the cycle. This is verified in simple time-series regressions of each of the poverty measures on the aggregate unemployment rate and a linear trend. The coefficient on unemployment falls from 0.78 (0.093) to 0.58 (0.069) to 0.46 (0.093) for market income to Census income to net income, respectively (standard error in parentheses). This suggests that the safety net is playing an important automatic stabilizer role.

Research Needs on Poverty and Social Policy

Much of the early poverty research focused on the disincentive effects of income transfers on work, saving, marriage, and related family outcomes (Danziger, Haveman, & Plotnick, 1981; Moffitt, 1992). This was an obvious starting point, both because economic theory often lent ambiguous predictions owing to potentially offsetting substitution and income effects and because the fiscal implications for government budgets were contingent on the magnitude of response (Barr, 1992). The general consensus from that literature was that while transfers discouraged some prosocial behaviors, the disincentive effects were anything but uniform, and generally small in magnitude.

More recently, attention has broadened beyond the costs of the safety net to examine the insurance and redistributive aspects of social assistance on eradicating poverty and hardship (Atkinson, 1999; Bartfeld, Gundersen, Smeeding, & Ziliak, 2015a; Blundell & Pistaferri, 2003; Currie, 2006; Duncan, Huston, & Weisner, 2007; Gruber & Dynarski, 1997; Hoynes, Schanzenbach, & Almond, 2016; Huggett & Parra, 2010; Kniesner & Ziliak, 2002; Kreider, Pepper, Gundersen, & Jolliffe, 2012; Waldfogel, 2011). It is too soon to claim consensus on this new research strand, but common themes are compelling evidence that the safety net is effective at reducing the magnitude and duration of shocks to household income and consumption, and that the programs have long-term positive effects on child welfare. Even though our understanding of how social policy affects family well-being has advanced greatly in recent decades, there still remain many open questions, some of which stem from conflicting evidence from prior research, while others have yet to be explored systematically. This section offers some suggested avenues for future research in the domains of poverty measurement and outcomes-based research dealing with program design. Again, while the institutional details are based on the United States, the general issues are not unique to the U.S. case.

Measurement

Some commentators have claimed that the War on Poverty in the United States was lost because poverty has been little changed. Indeed, the coefficient on the linear trend in a time series regression of market-income poverty and the unemployment rate depicted in Figure 1 is actually a positive 0.069 (0.015), meaning that pre-tax and transfer poverty is trending up, and the coefficient on the trend for the Census poverty rate model is 0.003 (0.011), meaning that there has been no progress in trend poverty. This is the news from the official poverty series, and thus the policymaker who only looks at the headline rate might be forgiven for making the claim of no progress. However, the claim is erroneous, as once one expands the definition to include in-kind transfers, tax credits, and tax payments there is robust evidence that substantial advances on reducing poverty have been made overall, and especially in rural areas of America (Nolan, Waldfogel, & Wimer, 2017). Indeed, the coefficient on the trend in the net income poverty rate regression is -0.099 (0.015), indicating a strong downward trend.

That the official poverty measure might provide a misleading assessment of the effectiveness of the safety net is well known, and this shortcoming formed the basis of the National Research Council’s report two decades ago arguing for a more inclusive measure of resources (Citro & Michael, 1995). The Census Bureau has taken the lead in this agenda in producing the Supplemental Poverty Measure (Renwick & Fox, 2016). Perhaps the greatest research challenge with broader measures like the SPM is how to properly value in-kind transfers (Moffitt, 1989; Smeeding, 1982). Some transfers like SNAP have been shown to be near cash, and thus a dollar of SNAP is like a dollar of cash. But other transfers like health care provided by Medicare and Medicaid fall far short of a dollar for dollar conversion because the benefit is generally redeemed only when the person is beset with illness. The question is how much below a dollar is a dollar of health insurance. Little work has been done on this topic for decades, but the need has increased dramatically. For example, in 1980 in-kind transfers outnumbered spending on cash assistance by a factor of roughly 2 to 1. Today that ratio exceeds 10 to 1. This has occurred both because of the rising cost of health care, but also from other programmatic changes in the safety net such as the TANF reform that resulted in the average state spending about 80% of their block grant on in-kind assistance and 20% on cash, which is just the opposite the former AFDC program. How such benefits are valued by welfare recipients has yet to be explored in any systematic fashion.

Besides in-kind transfers, there are (at least) two additional research challenges facing the resource side of poverty measurement, and evidence suggests the biases work in opposite directions. Poverty in most countries is measured from household surveys, and in the United States this is the CPS ASEC. One challenge with the CPS is rising rates of nonresponse in any aspect of the survey, in participation in the ASEC income supplement, and in responding to earnings questions. Bollinger, Hirsch, Hokayem, and Ziliak (2019) report that earnings nonresponse in the ASEC has more than doubled since 1990 to nearly 45% of the sample.5 The current practice at the Census is to replace that missing information with a so-called hot-deck imputation, which involves replacing the missing data with information from a randomly matched donor. The latter relies on the assumption that the data are missing at random, but this is rejected in the data, and as a result of this assumption, when linking the ASEC to administrative earnings records the official poverty rate in the United States is shown to be biased downward by about 10% (Hokayem et al., 2015). Research is needed on two fronts. First, we need work on how to improve survey participation to reduce the influence of imputation. Second, we need additional research on improving imputation methods for those who fail to respond. An argument can be made that the missing data should remain missing, that is, “just say no to imputation,” but it is also possible that there are alternative forms of imputation that may be more successful than current techniques are at reducing bias. Bollinger and colleagues (2019) show that there is some promise using quantile methods to correct for nonrandom sample selection, but more research is needed.

The second challenge with the CPS ASEC, and indeed in most household surveys in the United States, United Kingdom, and other countries, is deteriorating quality of transfer data from underreports of both participation and amounts (Brewer, Etheridge, & O’Dea, 2017; Meyer, Mok, & Sullivan, 2015). The evidence suggests that over the past couple of decades, households have been increasingly unwilling to report (or unable to recall) participation in transfer programs, and conditional on correctly claiming participation, reporting the correct dollar amount of assistance. This means estimates of the anti-poverty effectiveness of social policy in Figure 1 is understated. As with nonresponse, one area of research need is on how to better elicit recall to improve survey income reports. Another area, which is currently quite active, is how to integrate administrative data into survey reports of transfers.6 The challenge is that the purpose of state administrative systems is to monitor participation and program integrity, and not research per se. Thus the universe and file structure of administrative data and survey data are often distinct, and the quality of the administrative systems vary considerably across states. However, the promise of linking transfer program data directly with surveys will greatly improve our ability to evaluate programs and thus should be a priority research topic.

The innovation of the SPM is not only on the resource side, but also on the threshold side of poverty measurement. When the Orshansky (1963) threshold was designed, the typical family spent one-third of their net income on food purchases, but today this ratio is closer to one-eighth. This suggests that instead of inflating the food plan by a factor of three to draw the line, one should inflate it by a factor of eight. The NRC, however, proposed eschewing the Orshansky thresholds altogether and replacing them with a new line based on a broader measure of consumption (Citro & Michael, 1995). Census has implemented this in the SPM by summing up food, clothing, shelter, and utilities (FCSU), plus a 20% inflation factor on the latter sum to account for miscellaneous spending, and selecting the line as that level of spending at the 33rd-36th percentiles of FCSU. Ongoing research needs in this area include work on whether this is the correct bundle of goods to draw the line. For example, transportation is the second largest budget item in the typical consumer’s basket, but this is not included in the FCSU measure, and this could particularly bias against rural families who rely more on personal autos (Glaeser, 2011). In addition, the Orshansky line is fixed across the country under the assumption that there are no differences across geography in cost of living. Some SPM versions account for this by adjusting for differences in housing cost, but it is far from clear that this is sufficient and more work is needed.

Some have argued in favor of replacing the income-based measure of poverty in favor of a consumption-based approach (Meyer & Sullivan, 2012; Slesnick, 2001). The case is made on both theoretical and measurement grounds. On the theoretical side, the argument is that current income is only a snapshot of real purchasing power and at any given point in time it may be transitorily low. However, during periods of temporary income shortfalls the family may be able to maintain consumption by spending down assets and/or borrowing and thus current consumption better represents permanent income. Although this argument is compelling on the surface, Blundell and Preston (1996) urge caution because consumption is most credible as a measure of well-being when comparisons are made between households of a similar age and birth cohort because households of different cohorts may face a very different pattern of real interest rates that can affect the ability to intertemporally smooth consumption. On the measurement side, the argument rests heavily on the problems discussed above regarding underreporting of transfers, and nonresponse. In the United States, the only comprehensive dataset on consumption for poverty measurement is the Consumer Expenditure Survey (CE). However, some have criticized the CE because aggregate consumption in the CE does not align well with consumption reported in the National Income and Product Accounts (Attanasio, Hurst, & Pistaferri, 2014). More research is clearly needed on the relative merits of survey measures of income and consumption.

Program Design

Across scores of papers and decades, social scientists consistently provide evidence that the disincentive effects of social transfers to work, gain an education, and marry are small in magnitude, and thus reliance on assistance programs is primarily a consequence, and not a cause, of the poverty witnessed in recent decades (Bitler & Hoynes, 2016; Danziger, Haveman, & Plotnick, 1981; Moffitt, 1992; Ziliak, 2016). Despite this evidence, there remains general dissatisfaction among many across countries on the cost and effectiveness of the safety net. Recently there has been some important research highlighting how the safety net improves consumption and health outcomes, both contemporaneously and in the long-run for children who grew up in poverty (Almond et al., 2016; Atkinson, 1999; Bartfeld et al., 2015a; Blundell & Pistaferri, 2003; Currie, 2006; Duncan et al., 2007; Gruber & Dynarski, 1997; Huggett & Parra, 2010; Kniesner & Ziliak, 2002; Kreider et al., 2012; Waldfogel, 2011). Work along these lines is crucial to fill in the gaps of knowledge so that policy can be better informed by evidence. There are some new developments in the policy sphere that do not have adequate evidence to support one way or the other, and new research is needed.

One of these areas is the ascendance of work requirements. Policymakers in the United States and elsewhere continue to remake the safety net programs to favor those in-work versus those out-of-work. Indeed, unlike other nations in the OECD, the United States no longer has a general cash assistance program for those out of work but not disabled. The rise of the EITC in the early 1990s, coupled with the 1996 welfare reform, were the first major steps in this direction, requiring work to qualify for the EITC, and for those on TANF requiring most adult recipients to engage in work and to limit the amount of time on aid. The 1996 legislation expanded work requirements to able-bodied adults without dependents (ABAWDS) between the ages of 18 and 49 to qualify for SNAP—they must work at least 20 hours per week, or else they only qualify for 3 months of SNAP out of every 36-month period. Some members of Congress have proposed to expand these SNAP work requirements to childless persons up to age 60, and to able-bodied parents caring for children aged 6 and older. Several states are proposing to expand these work requirements to the Medicaid program, despite the possibility that the causation of non-work may run the reverse from the policymaker’s assumption, and the Trump administration is considering work requirements to qualify for housing assistance. What evidence does exist on the effect of work requirements mostly comes from the TANF program, or the demonstration projects that preceded them, and that work shows work requirements reduced welfare participation but did not substantially change (for the better or worse) other domains of life (Ziliak, 2016). However, this evidence base is thin and much more is needed given the prominence of work requirements in policy discussions. This push toward work is also the case in other countries. For example, the 2012 welfare reform in Great Britain proposed the Universal Credit to replace the working tax credit, child tax credit, housing benefit, income support, and other employment allowances.

Another area of research need on program design is the fiscal federalism of the welfare state. As described in the previous section, the current U.S. system is a hodgepodge of federal programs, federal-state matching programs, federal block grants, and sole state programs. With the Republican takeover of both houses of Congress and the election of President Trump in 2016, there were calls to reform the funding structure of welfare; namely, holding up TANF as a model, some proposed to block grant Medicaid, SNAP, SSI, and housing, with additional programs potentially added at a later date. There is a literature on the economics of the federal role in addressing poverty and on the relative effects and merits of block grants and matching rate grants (Baicker, 2005; Brown & Oates, 1987; Chernick, 1998; Gramlich, 1982; Marton & Wildasin, 2007; Pauly, 1973). However, with few exceptions, this work is focused almost exclusively on the pre-1996 AFDC program. There has been limited formal analysis of state spending under the TANF program, with its block grant and state-spending structure. The experience of TANF suggests that, left unconstrained, states will utilize the block grant structure to meet a variety of objectives different from those under a restricted matching grant which requires spending on specific activities. New research is needed to spell out the costs and benefits of restructuring the financial system of welfare assistance.

Finally, there is a push across many countries to replace some or all programs with a universal basic income (UBI). The concern is that the current system is administratively complex and costly (e.g., in the United States the safety net spans at least a half dozen federal agencies each with their own rules and regulations), that creates multiple implicit tax cliffs as eligibility changes across the income distribution and thus creating various work and saving disincentives, and as a result, many who are eligible to participate because of low-incomes and assets choose not to because of the hassle (Lowrey, 2018). The premise of UBI is built off the negative income tax (NIT) idea first proposed by Friedman (1962), whereby the only criterion to qualify for assistance is low incomes. The difference between UBI and the NIT is that the UBI would not be means-tested and would not be taxed away as income rose, that is, it is a per-person cash grant provided regardless of age, income, health, or family structure. As such, the UBI would only create an income effect in the work-leisure decision and thus there would be no substitution effect and thereby no deadweight loss compared to the current system or the NIT as envisioned by Friedman. There is little literature to guide policy, but a good starting point for new work in this area is the research by Saez (2002), Rothstein (2010) and others comparing the EITC and NIT. The introduction of Universal Credit in Great Britain, for example, offers opportunities to inform UBI ideas, at least to the extent one can test for efficiency gains from program consolidation.

Further Reading

Bartfeld, J., Gundersen, C., Smeeding, T., & Ziliak, J. (2015). SNAP matters: How food stamps affect health and well-being. Redwood City, CA: Stanford University Press.Find this resource:

Bitler, M., & Hoynes, H. (2010). The state of the social safety net in the post-welfare reform era, Brookings Papers on Economic Activity, 2, 71–127.Find this resource:

Blank, R. (1997). It takes a nation: A new agenda for fighting poverty. Princeton, NJ: Princeton University Press.Find this resource:

Cancian, M., & Danziger, S. (2009). Changing poverty, changing policies. New York, NY: Russell Sage Foundation.Find this resource:

Currie, J. (2006). The invisible safety net: Protecting the nation’s poor children and families. Princeton, NJ: Princeton University Press.Find this resource:

Grogger, J., & Karoly, L. (2005) Welfare reform: Effects of a decade of change. Cambridge, MA: Harvard University Press.Find this resource:

Haskins, R. (2007). Work over welfare: The inside story of the 1996 welfare reform law. Washington, DC: Brookings Institution Press.Find this resource:

Lowrey, A. (2018). Give people money: How a universal basic income would end poverty, revolutionize work and remake the world. New York, NY: Crown.Find this resource:

Meyer, B., & Sullivan, J. (2012). Identifying the disadvantaged: Official poverty, consumption poverty, and the new supplemental poverty measure. Journal of Economic Perspectives, 26, 111–136.Find this resource:

Moffitt, R. (2015). The deserving poor, the family, and the U.S. welfare system, Demography, 52, 729–749.Find this resource:

Moffitt, R. (2016). Economics of Means-Tested Transfer Programs in the United States, Volumes 1 and 2. Chicago, IL: University of Chicago Press.Find this resource:

Murray, C. (2006). In our hands: A plan to replace the welfare state. Washington, DC: AEI Press.Find this resource:

Ziliak, J. (2006). Understanding poverty rates and gaps: Concepts, trends, and challenges. Foundations and Trends in Microeconomics, 1, 127–199.Find this resource:

Ziliak, J. (2009). Welfare reform and its long term consequences for America’s poor. Cambridge, U.K.: Cambridge University Press.Find this resource:

References

Alkire, S., Foster, J., Seth, S., Santos, M., Roche, J., & Ballon, P. (2015). Multidimensional poverty measurement and analysis. Oxford, U.K.: Oxford University Press.Find this resource:

Anderson, L. (1964). Trickling down: The relationship between economic growth and the extent of poverty among American families. The Quarterly Journal of Economics, 78, 511–524.Find this resource:

Atkinson, A. (1987). On the measurement of poverty. Econometrica, 55, 749–764.Find this resource:

Atkinson, A. (1999). The economic consequences of rolling back the welfare state. Cambridge, MA: MIT Press.Find this resource:

Attanasio, O., Hurst, E., & Pistaferri, L. (2014). The evolution of income, consumption, and leisure inequality in the United States, 1980–2010. In C. Carroll, T. Crossley, and J. Sabelhaus (Eds.), Improving the measurement of consumer expenditures (pp. 100–140). Chicago, IL: University of Chicago Press.Find this resource:

Autor, D., & Duggan, M. (2006). The growth in social security disability rolls: A fiscal crisis unfolding. Journal of Economic Perspectives, 20, 71–96.Find this resource:

Baicker, K. (2005). Extensive or intensive generosity? The price and income effects of federal grants. Review of Economics and Statistics, 87, 371–384.Find this resource:

Banerjee, A., Benabou, R., & Mookherjee, D. (2006). Understanding poverty. Oxford, U.K.: Oxford University Press.Find this resource:

Barr, N. (1992). Economic theory and the welfare state: A survey and interpretation. Journal of Economic Literature, 30, 741–803.Find this resource:

Bartfeld, J., Gundersen, C., Smeeding, T., & Ziliak, J. (2015). SNAP matters: How food stamps affect health and well-being. Redwood City, CA: Stanford University Press.Find this resource:

Bee, C., Gathright, G., & Meyer, B. (2018). The determination of survey non-response bias through the use of tax records. Mimeo.Find this resource:

Bitler, M., & Hoynes, H. (2016). The more things change, the more they stay the same? The safety net and poverty in the great recession. Journal of Labor Economics, 34, S403–S444.Find this resource:

Bollinger, C., Hirsch, B., Hokayem, C., & Ziliak, J. (2019). Trouble in the tails? What we know about earnings nonresponse thirty years after Lillard, Smith, and Welch. Journal of Political Economy.Find this resource:

Blank, R., & Card, D. (1993). Poverty, income distribution, and growth: Are they still connected? Brookings Papers on Economic Activity, 2, 285–339.Find this resource:

Blundell, R., & Pistaferri, L. (2003). Income volatility and household consumption: The impact of food assistance programs. Journal of Human Resources, 38, 1032–1050.Find this resource:

Blundell, R., & Preston, I. (1996). Income, expenditure, and the living standards of UK households. Fiscal Studies, 16, 40–54.Find this resource:

Brewer, M., Etheridge, B., & O’Dea, C. (2017). Why are households that report the lowest income so well-off? The Economic Journal, 127, F24–F49.Find this resource:

Brown, C., & Oates, W. (1987). Assistance to the poor in a federal system. Journal of Public Economics, 32, 307–330.Find this resource:

Cancian, M., & Danziger, S. (2009). Changing poverty, changing policies, New York, NY: Russell Sage Foundation.Find this resource:

Chernick, H. (1998). Fiscal effects of block grants for the needy: An interpretation of the evidence. International Tax and Public Finance, 5, 205–233.Find this resource:

Citro, C., & Michael, R. (1995). Measuring poverty: A new approach, Washington, DC: National Academy Press.Find this resource:

Currie, J. (2006). The invisible safety net: Protecting the nation’s poor children and families. Princeton, NJ: Princeton University Press.Find this resource:

Danziger, S., Haveman, R., & Plotnick, R. (1981). How income transfer programs affect work, savings, and the income distribution: A critical review. Journal of Economic Literature, 19, 975–1028.Find this resource:

Duncan, G., Huston, A., & Weisner, T. (2007). Higher ground: New hope for the working poor and their children, New York, NY: Russell Sage Foundation.Find this resource:

Foster, J., Greer, J., & Thorbecke, E. (1984). A class of decomposable poverty measures. Econometrica, 52, 761–766.Find this resource:

Friedman, M. (1962). Capitalism and freedom. Chicago, IL: University of Chicago Press.Find this resource:

Glaeser, E. (2011). Measuring local poverty rates. In J. Ziliak (Ed.), Cost of living and the supplemental poverty measure, Submitted by the University of Kentucky Center for Poverty Research to the Office of the Assistant Secretary for Planning and Evaluation, U.S. Department of Health and Human Services.Find this resource:

Gottschalk, P., & Danziger, S. (1985). A framework for evaluating the effects of economic growth and transfers on poverty. The American Economic Review, 75, 153–161.Find this resource:

Gramlich, E. (1982). An econometric examination of the new federalism. Brookings Papers on Economic Activity, 2, 327–360.Find this resource:

Gruber, J., & Dynarski, S. (1997). Can families smooth variable earnings? Brookings Papers on Economic Activity, 1, 229–284.Find this resource:

Gundersen, C., & Ziliak, J. (2004). Poverty and macroeconomic performance across space, race, and family structure. Demography, 41, 61–86.Find this resource:

Hokayem, C., Bollinger, C., & Ziliak, J. (2015). The role of CPS nonresponse in the measurement of poverty. Journal of the American Statistical Association, 110, 935–945.Find this resource:

Hoynes, H., Schanzenbach, D., & Almond, D. (2016). Long-run impacts of childhood access to the safety net. The American Economic Review, 106, 903–934.Find this resource:

Huggett, M., & Parra, J. (2010). How well does the U.S. social insurance system provide social insurance? Journal of Political Economy, 118, 76–112.Find this resource:

Jefferson, P. (2012). The Oxford handbook of the economics of poverty, Oxford, U.K.: Oxford University Press.Find this resource:

Kniesner, T., & Ziliak, J. (2002). Explicit versus implicit income insurance. Journal of Risk and Uncertainty, 25, 5–20.Find this resource:

Kreider, B., Pepper, J., Gundersen, C., & Jolliffe, D. (2012). Identifying the effects of SNAP (Food Stamps) on child health outcomes when participation is endogenous and misreported. Journal of the American Statistical Association, 107, 958–975.Find this resource:

Kubik, J. (1999). Incentives for the identification and treatment of children with disabilities: The Supplemental Security Income Program. Journal of Public Economics, 73, 187–215.Find this resource:

Lowrey, A. (2018). Give people money: How a universal basic income would end poverty, revolutionize work and remake the world. New York, NY: Crown Publishing Group.Find this resource:

Marton, J., & Wildasin, D. (2007). State government cash and in-kind benefits: Intergovernmental fiscal transfers and cross-program substitution. Journal of Urban Economics, 61, 1–20.Find this resource:

Meyer, B., Mok, W., & Sullivan, J. (2015). Household surveys in crisis. Journal of Economic Perspectives, 29, 199–226.Find this resource:

Meyer, B., & Sullivan, J. (2012). Identifying the disadvantaged: Official poverty, consumption poverty, and the new supplemental poverty measure. Journal of Economic Perspectives, 26, 111–136.Find this resource:

Moffitt, R. (1989). Estimating the value of an in-kind transfer: The case of food stamps. Econometrica, 57, 385–409.Find this resource:

Moffitt, R. (1992). Incentive effects of the U.S. welfare system: A review. Journal of Economic Literature, 30, 1–61.Find this resource:

Moffitt, R. (2015). The deserving poor, the family, and the U.S. welfare system. Demography, 52, 729–749.Find this resource:

Nolan, L., Waldfogel, J., & Wimer, C. (2017). Long-term trends in rural and urban poverty: New insights using a historical supplemental poverty measure. In D. Lichter and J. Ziliak (Eds.), THE ANNALS: The New Rural-Urban Interface, 672, 123–142.Find this resource:

Orshansky, M. (1963). Children of the poor. Social Security Bulletin, 26, 3–13.Find this resource:

Pauly, M. (1973). Income redistribution as a local public good. Journal of Public Economics, 2, 35–58.Find this resource:

Renwick, T., & Fox, L. (2016). The supplemental poverty measure: 2015. Current Population Reports, P60–258, Washington, DC: U.S. Census Bureau.Find this resource:

Rothstein, J. (2010). Is the EITC as good as an NIT? Conditional cash transfers and tax incidence. American Economic Journal: Economic Policy, 2, 177–208.Find this resource:

Ruggles, P. (1990). Drawing the line: Alternative poverty measures and their implications for public policy, Washington, DC: Urban Institute.Find this resource:

Saez, E. (2002). Optimal income transfer programs: Intensive versus extensive labor supply responses. Quarterly Journal of Economics, 117, 1039–1073.Find this resource:

Sen, A. (1976). Poverty: An ordinal approach to measurement. Econometrica, 44, 219–231.Find this resource:

Slesnick, D. (2001). Consumption and social welfare: Living standards and their distribution in the United States. Cambridge, UK: Cambridge University Press.Find this resource:

Smeeding, T. (1982). Alternative methods for valuing selected in-kind transfer benefits and measuring their effect on poverty. U.S. Department of Commerce, Bureau of Census, Technical Paper 50.Find this resource:

Waldfogel, J. (2011). Britain’s war on poverty. New York, NY: Russell Sage Foundation.Find this resource:

Watts, H. (1968). An economic definition of poverty. In D.P. Moynihan (Ed.), On Understanding Poverty (pp. 316–329). New York, NY: Basic Books.Find this resource:

Zheng, B. (1997). Aggregate poverty measures. Journal of Economic Surveys, 11, 123–162.Find this resource:

Ziliak, J. (2006). Understanding poverty rates and gaps: Concepts, trends, and challenges. Foundations and Trends in Microeconomics, 1, 127–199.Find this resource:

Ziliak, J. (2015a). Recent developments in antipoverty policies in the United States. In J. K. Scholz, H. Moon, & S. Lee (Eds.), Social policies in an age of austerity (pp. 235–262). Cheltenham, U.K.: Edward Elgar Publishing.Find this resource:

Ziliak, J. (2015b). Why are so many Americans on food stamps? The role of the economy, policy, and demographics. In J. Bartfeld, C. Gundersen, T. Smeeding, & J. Ziliak (Eds.), SNAP matters: How food stamps affect health and well being (pp. 18–48). Redwood City, CA: Stanford University Press.Find this resource:

Ziliak, J. (2016). Temporary assistance for needy families. In R. Moffitt (Ed.), Economics of Means-Tested Transfer Programs in the United States, Volume 1 (pp. 303–393). Chicago, IL: University of Chicago Press.Find this resource:

Notes:

(1.) Research on the economics of poverty has exploded over the past five decades. For recent edited collections spanning both developed and developing economies (see Banerjee, Benabou, & Mookherjee, 2006; Cancian & Danziger, 2009; and Jefferson, 2012).

(2.) Sen (1976) argued that a good measure of poverty should satisfy at a minimum the focus, monotonicity, and transfer axioms. The focus axiom states that the measure should be relevant to the poor population, that is, transfers among the rich should have no bearing on the measure; the monotonicity axion states that the measure should rise and fall with changes in the resources of the poor person; and the transfer axiom states that a transfer from a poor person to a less poor person should result in increased poverty. The poverty rate only satisfies the focus axiom. The measure when a=1 satisfies the first two axioms, while a=2 satisfies all three. See Foster et al. (1984), Zheng (1997), and Ziliak (2006) for a discussion of these issues.

(3.) A similar critique could be levied against relative scales since one has to determine what fraction of income to draw the line, for example, 60% of equivalized median household income in Great Britain. There is nothing inherent that 60% of median income is the magic number capturing needs, let alone how one measures equivalence scales

(4.) The data come from the Current Population Survey Annual Social and Economic Supplement (ASEC) for survey years 1980–2014 (calendar years 1979–2013). The ASEC is a nationally representative survey of about 60,000 households conducted by the U.S. Census Bureau and serves as the source of official U.S. poverty and income statistics. Tax payments, along with the EITC and ACTC, are obtained by running each tax unit through the tax calculator provided by the NBER TAXSIM program.

(5.) This is the sum of ASEC supplement nonresponse and item nonresponse to the earnings question. Both forms of nonresponse are increasing, with item nonresponse accounting for about two-thirds of the total.

(6.) One legal challenge in the United States is that many of the social programs, while federally-funded, are controlled by the states who are under no mandate to share data with Census. Census is actively working on sharing agreements with the states.