Contents

• Introduction
• What Is Meant by "Poverty?"
• How Is Poverty Measured?
• Poverty Measures as Financial Measures Distinct from Other Measures of Well-Being
• Poverty Measures Are Not the Same as Sufficiency Measures
• Poverty Measures Focus on Income, Not Assets or Debt
• Poverty Measures Are Estimates Based on Survey Data
• The Official Poverty Measure
• Development of the Official Poverty Thresholds
• How the Official Measure Applies Poverty Measurement Concepts
• Comparing Resources and Needs: the Income-to-Poverty Ratio
• Whose Resources and Needs? Defining the Family Unit
• Limitations of the Official Poverty Measure
• The Supplemental Poverty Measure
• Thresholds
• Resources
• Comparison with the Official Measure
• Legal Authority for Poverty Measurement
• Data Sources Used in Poverty Measurement
• Current Population Survey (CPS) Annual Social and Economic Supplement (ASEC)
• Decennial Census Long Form
• American Community Survey (ACS)
• Small Area Income and Poverty Estimates (SAIPE)
• Small Area Health Insurance Estimates (SAHIE)
• The Survey of Income and Program Participation (SIPP) and the Survey of Program Dynamics (SPD)
• Department of Health and Human Services (HHS) Poverty Guidelines
• Simplification of the Poverty Thresholds for Administrative Use
• Development of the Guidelines
• Early Release of Inflation-Adjustment of the Guidelines to Support Programmatic Use

Summary

Poverty measures convey the number or percentage of people falling below given income amounts, which are intended to represent a level of economic privation and are computed using some factually based measurement of basic needs. The poverty measures discussed in this report—the official U.S. poverty measure and the research Supplemental Poverty Measure—focus on financial resources. A family's income is compared against a dollar amount representing some measure of need, called a threshold, which typically varies by family size and composition. Those with family income less than the threshold are considered to be "in poverty," or poor; those with incomes greater than or equal to the threshold are not considered to be in poverty. All members of the same family have the same poverty status.

The poverty measures discussed here are financial measures; they do not directly capture the physical, mental, or social effects of being poor. They were developed to accurately measure economic privation rather than to describe the full complement of resources a person or family needs to be self-sufficient.

Poverty data are obtained from surveys, and are therefore estimates that have margins of error. Poverty estimates derived from different data sources—even those using the same definition of poverty—will almost always differ.

The official poverty thresholds were developed in the early 1960s, and were based on empirical measures of dietary need, on the amount that a family in economic distress might need to spend on food to attempt to meet its dietary needs, and on the spending patterns of families across the income distribution. This information was used to determine what percentage of an average family's budget was spent on food, and in turn, to compute the amounts representing total family income.

There has been broad agreement among poverty scholars that the official poverty measure has serious limitations, and decades of research were undertaken to address them. In 2009, an interagency technical working group, convened under the auspices of the Office of Management and Budget (OMB), put forth the Supplemental Poverty Measure to consolidate the research and emphasize not only sound concepts and methodology in the measure's development, but also practicality in the measure's maintenance, computation, and usage. The Supplemental Poverty Measure was not intended to replace the official measure, and it was expected that refinement of the Supplemental Poverty Measure's methodology and data sources would continue.

Neither the official poverty measure nor the Supplemental Poverty Measure was established in statute. The Bureau of the Budget and its successor agency, OMB, directed federal agencies to use the official measure for statistical purposes. The directive explicitly stated that the measure was not developed for administrative purposes, and allowed for other measures of poverty to be developed, as long as the data for those measures were distinguished from the official series.

For administrative uses, such as determining whether an individual or family is eligible for assistance from a program, a different set of dollar amounts called poverty guidelines is used. Poverty guidelines are different from the official poverty thresholds, are published by the Department of Health and Human Services, and are not used to count the poverty population. However, any program that relies on counts of the poverty population, such as for formula grants, uses the official poverty thresholds and not the guidelines.

---

Introduction

Poverty estimates—the number and percentage of persons living in poverty—have been of interest to Congress not only to gauge the nation's economic health, but also because they are used to determine funding allocations for a variety of programs. Understanding how poverty is defined, how it is measured, and what information is used to compute poverty estimates can help congressional policymakers more adeptly consider how proposed legislation might affect the low-income population. To that end, this report provides a synopsis of poverty measurement in the United States, focusing on the following:

• the official measure of poverty, which is used to obtain official counts and percentages of the poor;
• the Supplemental Poverty Measure, which is used for research purposes only and was developed to improve upon some limitations of the official measure; and
• the Health and Human Services (HHS) poverty guidelines, which are used in administering programs for low-income persons but not to measure the poor population.

This report discusses poverty measures developed for official, research, and administrative uses in the United States. It does not address international poverty measures, current poverty estimates, or policy responses for alleviating poverty.¹

What Is Meant by "Poverty?"

The American Heritage Dictionary of the English Language defines poverty as "the state of being poor; lack of the means of providing material needs or comforts." "Poor" in turn is defined as "having insufficient wealth to meet the necessities or comforts of life or to live in a manner considered acceptable in a society."² Central to these definitions are "needs," "comforts," and living "in a manner considered acceptable in a society." Taken together, they characterize poverty as an economic concept that refers to limited resources: being in poverty indicates a level of hardship or deprivation because resources are insufficient to meet a level of material well-being. In the United States, economic resources often are expressed in monetary terms, and that expression makes poverty a financial concept. Persons living in poverty lack the financial resources to meet a basic level of material well-being.

How Is Poverty Measured?

Poverty measures convey the number or percentage of people falling below given income amounts, which are intended to represent a level of economic privation and are computed using some factually based measurement of basic needs. The poverty measures discussed in this report focus on financial resources. A family's income is compared against a dollar amount representing some measure of need, called a threshold, which typically varies by a family's size and composition. Those with family income less than the threshold are considered to be "in poverty," or poor; those with incomes greater than or equal to the threshold are not considered to be in poverty. All members of the same family have the same poverty status.

Poverty Measures as Financial Measures Distinct from Other Measures of Well-Being

The poverty measures discussed in this report focus entirely on economic needs and resources. This is only one way to assess a person's level of well-being, and other characteristics are also important to consider.

Some extended measures of well-being have been gauged through surveys. ³ These include the following:

• material well-being, which can be measured by the possession of amenities or access to public services;
• access to quality housing;
• neighborhood conditions and community services, such as public safety and quality local schools; and
• "the expectation of help, should need arise, from family, friends, or the community."⁴

Other indicators of well-being include economic mobility (the degree to which persons at the low end of the income distribution can move higher in it), health outcomes and risk factors,⁵ or access to opportunities for high paying jobs and geographic access to those jobs.⁶

These indicators of well-being tend to be inversely related to poverty, but the poverty measures discussed here were not designed to address them directly. They are financial measures, and do not directly capture the physical, mental, or social effects of being poor. That said, financial resources are nevertheless integral to well-being, and to the extent that poverty represents an inability to purchase or provide for a basic level of needs, these poverty measures do illustrate economic privation.

Poverty Measures Are Not the Same as Sufficiency Measures

Even within their focus on financial resources, the purpose of these poverty measures when they were developed was to accurately measure economic privation rather than to describe the full complement of resources a person or family needs to be self-sufficient. Mollie Orshansky, the researcher whose work formed the basis of the official poverty measure, characterized the distinction this way: "if it is not possible to state unequivocally 'how much is enough,' it should be possible to assert with confidence how much, on an average, is too little."⁷

Poverty Measures Focus on Income, Not Assets or Debt

While the American Heritage Dictionary's entry for "poor" referred to an absence of wealth, the poverty measures used in the United States focus on the flow of financial resources (income) received by families and individuals, rather than on assets or debt (which are the typical measures of wealth), as the relevant resources for meeting their needs. The focus on income rather than assets was driven at first by the available data when the official poverty measure was developed—the best data source available for poverty measurement did not ask about assets or debt. Furthermore, not all assets (assuming a person or family has any) can be readily liquidated in order to provide for basic needs, and the ones that are (e.g., savings or credit) tend to be short-term.⁸ The poverty measures discussed in this report are annual measures, which were not designed to focus on families and individuals that experience a short-term loss of income.

Poverty Measures Are Estimates Based on Survey Data

Poverty data are obtained from surveys, and are therefore estimates that have margins of error. Poverty estimates derived from different data sources—even those using the same definition of poverty—will almost always differ.

The Official Poverty Measure

Poverty estimates are obtained using household surveys. The Census Bureau releases data from the official poverty measure every September for the previous full calendar year, based on the Current Population Survey Annual Social and Economic Supplement (CPS ASEC).⁹ "Resources" in this measure are defined as money income before taxes, and they do not include the valuation of noncash benefits like the Supplemental Nutrition Assistance Program (SNAP) and housing subsidies. As its measure of need, it uses 48 dollar amounts, called poverty thresholds, which vary by family size and the members' ages. These thresholds are updated for inflation using the Consumer Price Index for All Urban Consumers (CPI-U), but are not adjusted geographically—the same thresholds are used throughout all 50 states. Both income and the thresholds in the official measure refer to the previous full calendar year. Table 1 displays the poverty thresholds for 2015.

As shown above, the thresholds do not increase by the same dollar amount between each family size.¹⁰ Similarly, within each family size, the thresholds do not increase or decrease by the same amount as the number of children increases and the number of adults decreases. The manner in which the threshold dollar amounts vary across family types and compositions is called the equivalence scale. Equivalence scales capture economies of scale in maintaining families. Some costs needed to maintain a family are fixed and do not change with family size, while others are directly affected by the number of family members. The official measure's equivalence scale was not computed directly, but rather came about implicitly because the thresholds were based on food costs and data on family expenditures.

Development of the Official Poverty Thresholds

With the launch of the War on Poverty during the Johnson Administration, the federal government devoted attention toward poverty measurement. The 1964 Economic Report of the President, in Chapter 2, "The Problem of Poverty in America," used a flat figure of $3,000 for families, and $1,500 for individuals not in families, to represent the poverty line. However, research on poverty measurement using a more detailed measure of need began before the Johnson Administration announced the War on Poverty. In the early 1960s, Mollie Orshansky, an analyst at the Social Security Administration, had been conducting her own analysis of poverty using food costs.¹¹ Rather than attempting to compute the costs of all major goods and services a family would need to live, the Orshansky thresholds were based on the Economy Food Plan, developed by the U.S. Department of Agriculture (USDA). The Economy Food Plan indicated the money that a family under economic stress would need to spend on food for "temporary or emergency use when funds are low."¹² The food plan used different amounts of food (and in turn, different costs of food) for males and females by age group, based on recommended dietary allowances.¹³ Orshansky condensed the full detail by specific age groups and for males and females into amounts that varied by family size, number of children under 18, and whether the family head was older than 65. Having obtained the cost of food by family size and composition, Orshansky then derived the amounts of total family income by finding the average percentage of it that families spent on food. According to the USDA's 1955 Household Food Consumption Survey, families spent an average of one-third of their income on food; therefore, multiplying the family food costs by three provided figures for total family income. For two-person families, a factor of 3.7, rather than 3, was used in order to capture the higher fixed costs faced by smaller families. Income amounts for individuals not in families were set at 80% of the amounts for two-person families.

Apart from three minor methodological revisions, the above methodology undergirds the official poverty thresholds as they are used today. ¹⁴ They are updated for inflation annually using the Consumer Price Index for All Urban Consumers (CPI-U).

How the Official Measure Applies Poverty Measurement Concepts

Poverty measurement, in general, involves a comparison of resources to a measure of need. Differences among poverty measures involve variations in the definitions of resources or needs, or both.¹⁵ "Resources" and "needs" typically refer to a family unit—the resources of all family members (however the members are identified) and the level of need applicable to that family and all its members. The definition of "family" also differs across measures, in terms of who counts as being "related" to whom.

Comparing Resources and Needs: the Income-to-Poverty Ratio

The comparison between resources and needs is usually expressed as a ratio: resources / needs. A ratio of less than one—meaning resources are less than needs—indicates the person or family is in poverty; a ratio greater than or equal to one indicates the person or family is not in poverty. This same ratio, when expressed as a percentage, conveys a measure of the severity of poverty. A ratio of exactly one is the same as "100% of the poverty level," and is the minimum amount to be considered "not in poverty." A ratio of less than 50% (or 0.5) has been characterized as "severe poverty" or "deep poverty" and indicates that resources are less than half the measure of need. Other ratios, such as 200% of poverty (twice the poverty level), have been used for measuring the low-income population intended to be the target beneficiaries of some programs, such as the State Children's Health Insurance Program (CHIP). The higher the ratio is, the higher the income amount will be, and therefore more people will fall below that amount. These ratios, called "income-to-poverty ratios," convey the sense that economic well-being is not binary (poor versus not poor), but rather it falls along a spectrum.

Whose Resources and Needs? Defining the Family Unit

In the poverty measures discussed in this report, resources and needs are defined for family units. The information used to identify family relationships comes from household surveys, and varies from survey to survey. The CPS ASEC, which is used for the official poverty estimates for the nation as a whole, uses a detailed questionnaire administered by trained representatives, and asks about the relationships among all household members.¹⁶

Because the information comes from household surveys, there are limitations to which populations can be measured. The CPS ASEC is conducted primarily for households, but also includes some noninstitutional group quarters, such as emergency shelters and group homes. Persons who are homeless are only included in the poverty estimates if they are living in shelters. Institutional group quarters, such as prisons and nursing homes, are not included, and therefore neither the incarcerated population nor those in other institutions are included in poverty estimates. College students living in dormitories are only included in poverty estimates if information about them is reported in an interview at their parents' home. Students living in households off-campus, however, are included. Military personnel are included only if they live in a household with at least one civilian adult.

In the official poverty measure, people related to each other by birth, marriage, or adoption are considered to be related. Foster children under age 15 who have not been legally adopted are excluded. In the CPS ASEC, children under age 15 are not asked questions about income; thus, income is unknown for any child under 15 whose family relationship is not tied to an older relative in the household whose income is known. The unrelated individuals under age 15 (mostly foster children) are not considered to be "poor" or "nonpoor," and their removal from the total prevents them from having an effect on poverty rates (the percentage of the total that is poor). Unmarried domestic partners are regarded as belonging to separate family units in the official definition of poverty, and their poverty statuses are computed accordingly.¹⁷

It is therefore important in measuring poverty that the data source provide information to identify family relationships—who is related, and who is not—and can provide a means to match income amounts with specific family units. The family relationship data available through the survey can thus affect how individuals' poverty status is determined, as does the definition of poverty used.

Limitations of the Official Poverty Measure

The official poverty measure represented an improvement from the flat amounts used in the 1964 Economic Report of the President, but it has nonetheless received criticism for decades:¹⁸

• 1. The poverty thresholds are not adjusted to reflect geographic variations in costs.
• 2. Owing to the limitations of the source data available at the time the official measure was developed, its resource measure is based on money income before taxes, while most individuals pay for their basic necessities using after-tax income. This represents an important disconnection between the way needs were specified in the thresholds and the definition of resources available for meeting those needs.
• 3. The official measure captures the effects of some but not all government programs intended to provide relief for the poor. The programs that are captured are those based on money income before taxes: Social Security, Supplemental Security Income (SSI), Temporary Assistance for Needy Families (TANF), and any state or local relief programs based on money income. The programs that are not captured are the Earned Income Tax Credit (EITC), which, despite its large effects for low-income workers,¹⁹ is not captured because it is a tax credit and is only reflected in after-tax income; and a host of noncash benefits such as SNAP, the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC), public housing subsidies, and subsidized medical care.
• 4. The official measure does not capture the needs incurred, nor the resources brought in, by household members who are not related by birth, marriage, or adoption; these include unmarried partners and their children (if any are present), and foster children not legally adopted.
• 5. While the official measure is adjusted for overall inflation, it does not adequately consider the extent to which the prices of basic necessities have shifted in relation to all goods and services. Therefore, it can be argued that the inflation adjustment used in the official measure does not accurately reflect the purchasing power needed, in a practical sense, to remain comparable with those at the poverty line in previous decades.

The Supplemental Poverty Measure

There has been broad agreement among poverty scholars that the limitations noted above represent serious drawbacks to the official poverty measure, and also that fixing them has not been a straightforward task. Experts within the federal government, in universities, and in private research institutions have spent decades developing approaches to address these shortcomings, and evaluating the effectiveness of those approaches.²⁰ In 2009, the Office of Management and Budget (OMB) convened an Interagency Technical Working Group (ITWG) to consolidate the research and propose a single Supplemental Poverty Measure, which would be used for research purposes only. The ITWG did not recommend supplanting the official poverty measure, nor did it recommend using the Supplemental Poverty Measure for administrative purposes. Comments on the Supplemental Poverty Measure were solicited through a Federal Register notice.²¹ Pursuant to research recommendations, the Supplemental Poverty Measure focused on ensuring that the definition of need specified in the thresholds, and the resources available to meet those needs, were consistent with each other. Additionally, the measure's methodology was designed to be practical to implement, with the understanding that its methods would continue to be refined periodically, as new data source availability and research findings warranted.

Thresholds

Following recommendations made by a National Academy of Sciences (NAS) panel, the thresholds were defined by costs of food, clothing, shelter, and utilities, and a bit extra (20% of the other costs) to cover miscellaneous expenses, as measured by the Consumer Expenditure Survey (CE). Three sets of thresholds were developed—those for renters, homeowners with a mortgage, and homeowners without a mortgage—to take account of housing costs, given that it can vary widely between these groups. The thresholds are geographically adjusted using housing data from the American Community Survey (ACS), because its large sample size enables it to provide detail for areas smaller than states. The equivalence scale for the new thresholds uses three factors to adjust for family size: (1) children on average consume less than adults; (2) as family size increases, some expenses like food or clothing increase additively with each person but other fixed expenses like housing do not; and (3) the first child in a single-parent family represents a greater increase in costs than the first child in a two-adult family. The Bureau of Labor Statistics (BLS) computes the base thresholds using CE data; ACS data are used in geographic adjustment.²² Table 2 illustrates a few of these thresholds from each census region of the nation (South, Northeast, Midwest, and West).

Resources

Income in the Supplemental Poverty Measure was defined to represent resources available to meet basic needs. It was based on after-tax income, and includes the effects of EITC. The values of noncash benefits were estimated, to reflect resources available for meeting the food and housing costs specified in the thresholds and to capture the effect of government programs intended to help the poor meet those needs. Additionally, work-related expenses were subtracted from income, as the money that a person applies toward his or her job cannot be used to meet the needs in the threshold. These expenses include transportation to and from work, child care (capped at the amount earned by working), and any miscellaneous costs that workers are required to pay as part of their job. Medical out-of-pocket expenses were also subtracted from income, such as co-pays at doctor's offices, insurance premiums, and deductibles. This approach was intended to capture that medical costs in many cases are not optional and must be paid in order for basic needs to be met—and therefore money applied toward those costs is not available for meeting other basic needs—but also that a person who incurs a large amount of health care costs that are paid for by government or private insurance should not be viewed as wealthier than a healthy person with otherwise identical needs and resources. As a result, the Supplemental Poverty Measure does not capture the full effects of Medicare and Medicaid on the poor population, because the health care needs that those programs address are not the same as the economic needs that are incurred on a recurring and predictable basis specified in the thresholds (i.e., food, clothing, shelter, and utilities). The resource definition used in the Supplemental Poverty Measure was developed to be consistent with the measure of need as it was specified in the thresholds.

Comparison with the Official Measure

Annual reports and data files using the Supplemental Poverty Measure are produced by the Census Bureau, using income data from the CPS ASEC and the poverty thresholds provided by BLS. The Supplemental Poverty Measure's overall effect on the poverty rate (i.e., the percentage of people with income below the thresholds) made it approximately 0.6 percentage points higher than the official rate in 2015. However, the profile of the population in poverty is different between the supplemental and official measures, as evidenced by noticeably higher poverty rates for some groups (such as working-age adults and people 65 and older) and lower rates for others (such as children).²³

Legal Authority for Poverty Measurement

Neither the official poverty measure nor the Supplemental Poverty Measure was established in statute. The official measure gained its status through the Bureau of the Budget's Circular A-46, issued in 1969. OMB, the successor agency to the Bureau of the Budget, issued Statistical Policy Directive 14 in 1978, reconfirming the measure as official and directing federal agencies to use it for statistical purposes.²⁴ The directive explicitly stated that the measure was not developed for administrative purposes, and allowed for other measures of poverty to be developed, as long as the data for those measures were distinguished from the official series.

Given that federal agencies, such as the Census Bureau, were directed to use the official measure for statistical purposes (such as obtaining the number of people in poverty), any legislation that requires a count of the number of people in poverty (or a count below some other ratio of income to poverty) in order to determine formula grants makes use of the official poverty measure, even if the text of the legislation does not name the measure explicitly. Two examples are the state formula grant for the State Children's Health Insurance Program (CHIP), and Title I-A funding allocation formulas for school districts. A more complete list of programs is available in Appendix D of CRS Report R43863, Federal Benefits and Services for People with Low Income: Programs and Spending, FY2008-FY2013, by Karen Spar and Gene Falk.

Like the official poverty measure, the research Supplemental Poverty Measure was not established in statute.²⁵ It was developed after decades of research both within and outside the federal government. Leading up to that effort, dozens of alternative poverty measures had been developed, some as early as the 1980s. OMB provided the auspices for an ITWG that recommended the methodology for the Supplemental Poverty Measure, and solicited public feedback through a Federal Register notice.

Data Sources Used in Poverty Measurement

Current Population Survey (CPS) Annual Social and Economic Supplement (ASEC)

The CPS ASEC is the official source of annual estimates of poverty for the nation, and the Census Bureau continues to release poverty data every September from the ASEC for the prior full calendar year. The CPS is a survey sponsored by BLS and collected by the Census Bureau, and is used by BLS to report the monthly unemployment rate. At the end of the basic monthly labor force survey, an additional set of questions is asked (a supplement). Each month's supplement focuses on a different topic, so as to obtain additional information without having to conduct an entirely separate survey with its own costs of sample design, collection, and data processing. Every March, the extra questions focus on income in the previous year, with detail on 19 different income sources, as well as other topics such as health insurance coverage in the previous year, and receipt of noncash benefits (collectively these extra questions make up the ASEC). Because the CPS is administered using professional interviewers with computerized questionnaires, the CPS ASEC is able to gather more detail from respondents with greater accuracy than other surveys that rely on the respondents to fill out paper questionnaires. At the same time, using professional interviewers means that the cost per response is greater than with questionnaires filled out by the respondents, and thus the sample size is not as large—which in turn means that while the CPS ASEC offers greater detail at the national level, estimates for states and smaller areas are not as statistically reliable.

The official poverty measure, as described in OMB's Statistical Policy Directive 14, is based on data from the CPS ASEC.²⁶ Compared with other data sources available in the 1960s, when the official poverty measure was first computed, the CPS offered several advantages that made it the only reasonable choice for poverty measurement. First, it is an annual survey that asks detailed questions about income in the previous calendar year. This frequency of data collection meant that national poverty estimates could be released annually as opposed to once every 10 years, as was the case for the decennial census. Second, while there were other sources of annual data on consumer income, the CPS was the only nationally representative source with sufficient detail on both income and family characteristics that could feasibly work with Mollie Orshansky's approach to poverty measurement ²⁷—computing family income using some empirical basis of need—which was eventually adopted as the official approach.

While the CPS ASEC is the source of official poverty data for the nation, its relatively small sample size (approximately 100,000 addresses nationally) means that the margins of error are wide for state poverty estimates, and, with the exception of the largest metropolitan areas, too wide to produce meaningful estimates for counties and other small geographic areas—in fact, not every county or city is in the sample every year.²⁸ Since the 1960s, however, other datasets have been used to compute poverty status.

Decennial Census Long Form

The Decennial Census of Population and Housing was, for decades, the only available data source that could provide information on poverty for small geographic areas, such as cities and counties, and could even provide poverty data for areas as small as school districts and census tracts.²⁹ The decennial census poverty data were obtained from information collected on the "long form," a detailed questionnaire that was sent to one out of every six households. While technically still a sample, the long form sample size collected enough responses from small areas that the estimates from the sample were meaningful. Except for some differences in the data collection methods, the amount of income detail collected, and the ways families were defined, the same overall definition of poverty was used in the decennial census as was used in the CPS.³⁰ The same dollar amounts were used as poverty thresholds, and income was generally defined the same way. However, poverty data from the decennial census were only available every 10 years, for the full calendar year preceding the census: 1959, 1969, 1979, 1989, and 1999.

American Community Survey (ACS)

The long form was discontinued in Census 2010. In its place, the ACS collects information similar to what was collected on the long form, but instead of surveying one-sixth of all households every 10 years, the ACS collects information every month. By combining five years' worth of data, the ACS provides a sample size similar to the long form, with the same detail for small geographic areas, but on a recurring basis. The Census Bureau produces poverty estimates from the ACS each year, using both one-year estimates (for places with populations of 65,000 or more) and five-year estimates (which include all geographic areas, including those with populations below 65,000). While the ACS was designed to be filled out by the respondent on his or her own, and therefore asks about fewer income sources than does the CPS, its much larger sample size, coupled with the wide array of topics covered by the questionnaire, means that it currently provides more detail on demographic characteristics for small areas than any other data source.

Respondents are asked to provide each person's relationship to the householder in the ACS. As a result, it does not capture information on people who may be related to each other but not to the householder. The ACS treats these persons as unrelated individuals, and their poverty status may be computed differently than if their relationship to each other were known.

Small Area Income and Poverty Estimates (SAIPE)

Title I-A of the Elementary and Secondary Education Act (ESEA) establishes formula grants for local education agencies based, in part, on the number of school-age children in poverty within the school district.³¹ Because the formula relies on the count of the poor population for an area, the official poverty measure based on the Census Bureau thresholds is used, not the HHS poverty guidelines (discussed below). Before the advent of the SAIPE program, local officials had to wait 10 years before they could get meaningful poverty estimates for their area. The SAIPE program was undertaken in the 1990s (the first dataset, for 1993, was released in 1997) to provide poverty estimates for counties, and estimates of the number of school-age children in poverty for school districts, in between the decennial census data releases.

The SAIPE data are model-based, meaning that they are not simply a reflection of survey responses, but are adjusted using a combination of administrative data in addition to survey data. The modeling enables the Census Bureau to produce estimates for small areas with less variability—and thus greater precision—than it could using survey data alone.

Initially, SAIPE estimates were based on CPS data in addition to administrative data. Since 2005, the first year that the ACS estimates were available at its full sample size of over 3 million addresses per year, ACS data have been used as input into the model, along with anonymous IRS tax returns, SNAP administrative data, and the Census Bureau's own annual population estimates.

While small area estimates are produced annually from the ACS using five consecutive years' worth of data, the SAIPE model-based estimates reflect the prior year alone.³² Because SAIPE estimates are modeled to reflect a single year, they represent current economic conditions at the local level more accurately than the ACS five-year estimates. The SAIPE estimates, however, are limited in focus—they provide counts and percentages (poverty rates) for the total population and for children, and estimates of median income, but are not designed to provide highly detailed profiles of the population for small areas.

Small Area Health Insurance Estimates (SAHIE)

Like SAIPE, SAHIE data are also model-based estimates for small areas; however, they have a different target population: the low-income uninsured. The SAHIE estimates are broken down by income-to-poverty ratio, uninsured status, sex, age group, and, for state-level estimates, selected racial and Hispanic origin groups. One of the uses of SAHIE data is to estimate the target population for the National Breast and Cervical Cancer Early Detection Program, coordinated by the Centers for Disease Control and Prevention (CDC). In that program, uninsured women under selected income-to-poverty ratios and age groups qualify for subsidized mammograms and Papanicolaou tests. To be able to effectively monitor take-up rates of the program, the CDC provides the Census Bureau with some funding to support SAHIE. The SAHIE estimates also have been used to estimate the eligible population that would qualify for certain assistance under the Affordable Care Act: namely, in states that participate in the Medicaid expansion, the population with income below 138% of the poverty level; additionally, SAHIE provides estimates of uninsured children below 400% of poverty.³³

The Survey of Income and Program Participation (SIPP) and the Survey of Program Dynamics (SPD)

The data sources discussed thus far can be thought of as snapshots illustrating the poverty population in a given year.³⁴ While they are used to observe changes in the poverty population at an aggregate level, they usually do not interview the same respondents in consecutive years. As a result, they are not well-suited to measure the changes in a particular respondent's poverty status over time—or for that matter, changes to other social or economic conditions that he or she may experience such as changes in marital status, changes in the number of children in his or her family, obtaining or losing work, changes in his or her health insurance coverage or health status, receiving or no longer receiving government assistance, changing residence, etc.

A longitudinal survey is one that follows the same respondents over a series of interviews. Longitudinal poverty data have been available from the SIPP since 1984. The length of the time over which respondents have been interviewed has varied through the years, and has usually lasted from two to four years. By using SIPP longitudinal data, researchers have been able to identify characteristics of both the short-term and long-term poverty populations, the percentage of people who enter or leave poverty, and the average number of months a person stays in poverty.

In order to measure the initial effects of welfare reform under the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (PRWORA), longitudinal data were needed for the same respondents from both before and after welfare reform took effect. The SPD was created by selecting households from the 1992 and 1993 SIPP Panels for an additional five years of interviews, from 1997 to 2002.³⁵

The households that were in sample before the passage of PRWORA in 1996 and chosen for the SPD are no longer being interviewed, and in that sense the SPD has concluded. However, interest in the dynamics of poverty and program participation and in evaluating welfare reform continues, and is funded through mandatory spending in the Health and Human Services budget. Since the end of the original SPD panel, these dynamics have been and continue to be measured in the SIPP.

Department of Health and Human Services (HHS) Poverty Guidelines

Shown in Table 3 are sets of dollar amounts, called poverty guidelines, that are computed and published annually by the Department of Health and Human Services. These are different from the Census Bureau's poverty thresholds shown in Table 1. Unlike the measures discussed in this report, the poverty guidelines are not used to count the number of people or families in poverty. Rather, they are used for administrative purposes, such as determining whether an individual or a family qualifies for programmatic assistance.

Programs sometimes use different multiples of the poverty guidelines for determining eligibility. For instance, under the National School Lunch Program, a child qualifies for free lunch if his or her household's income is less than 130% of (that is, 1.3 times) the poverty guidelines, and qualifies for a reduced-price lunch if his or her household income is 185% of (that is, 1.85 times) the poverty guidelines.³⁶ These poverty guidelines are referenced in the Omnibus Budget Reconciliation Act of 1981.³⁷

The HHS poverty guidelines are not used to measure the poverty population, and therefore this report does not refer to them as a poverty "measure." Because both the poverty guidelines and the poverty thresholds are issued by federal agencies, either set of dollar amounts could conceivably be called the "federal poverty level" (FPL), and thus that term is ambiguous.³⁸ Which dollar amounts are to be used depends on the context: thresholds for obtaining counts of the population in poverty; or guidelines for administrative use, particularly with regard to an individual's or family's program eligibility. To clarify, the official poverty measure discussed in this report is based on the poverty thresholds that are computed and published by the Census Bureau, and are used for statistical purposes.

Simplification of the Poverty Thresholds for Administrative Use

As explained earlier, the official poverty thresholds were based on empirical measures of dietary need, the amount that a family in economic distress might need to spend on food in order to attempt to meet its dietary needs, and the spending patterns of families across the income distribution (to determine what percentage of an average family's budget was spent on food). However, this method yielded dozens of thresholds that were not evenly stepped by family size. As the War on Poverty developed, a set of dollar amounts was needed for practical administrative use. The HHS poverty guidelines serve that purpose.

Development of the Guidelines

The HHS poverty guidelines do not directly reflect family needs as measured by food cost and family spending data because they were simplified for administrative use. Beginning in 1965, the Office of Economic Opportunity (OEO) began to issue poverty guidelines that were based on the poverty thresholds, but with one dollar amount per family size (and at that time, farm/non-farm status).³⁹ These earliest OEO dollar amounts, which were the predecessors of the HHS poverty guidelines, were averages of the poverty thresholds that Orshansky developed. To obtain one dollar amount per family size, OEO averaged the thresholds that had the same family size but different compositions, weighted the average to reflect that not all family compositions occurred with equal frequency, and rounded to the nearest $5. A 1967 OEO memo introduced guidelines that were rounded to the nearest $100 and adjusted so that family size increments were evenly spaced by the same dollar amount. Like that of OEO, current HHS practice also applies rounding, but instead of rounding to the nearest $100, the difference between family sizes is computed first, rounded to the nearest $20, and then used to compute the guidelines for each family size, starting from a rounded figure for a four-person family as the base.⁴⁰

Early Release of Inflation-Adjustment of the Guidelines to Support Programmatic Use

The HHS poverty guidelines are published every winter, typically in late January or early February. They are computed using the weighted average thresholds published by the Census Bureau in its annual poverty release the previous fall, inflated forward to the current year using the Consumer Price Index. The steps to even out the differences between family sizes, and apply rounding, are performed after the inflation adjustment. By applying the inflation adjustment to the poverty thresholds from the previous year, HHS is able to publish the guidelines for the current year six to eight months before the Census Bureau releases that year's survey estimates, and thus they can be used by program administrators that much sooner. Table 3 provides the 2017 poverty guidelines.

Author Contact Information

Footnotes

1.	International data on poverty are available from the World Bank at http://www.worldbank.org/en/topic/poverty. For a concise framing of U.S. poverty trends, policy, and current considerations, see CRS In Focus IF10562, Poverty and Economic Opportunity, by Gene Falk, Maggie McCarty, and Joseph Dalaker. For specifics on programs that alleviate poverty, see CRS Report R43863, Federal Benefits and Services for People with Low Income: Programs and Spending, FY2008-FY2013, by Karen Spar and Gene Falk. For an overview of the U.S. poverty measure with recent data and trends, see CRS Report R44644, Poverty in the United States in 2015: In Brief, by Joseph Dalaker. For an examination of the population living below poverty, see CRS Report R44698, Demographic and Social Characteristics of Persons in Poverty: 2015, by Joseph Dalaker, Gene Falk, and Maggie McCarty.
2.	The American Heritage Dictionary of the English Language, Fifth Edition, Houghton Mifflin Harcourt Publishing Company, 2016.
3.	Reports and tabulations regarding the extended measures of well-being bulleted here are available on the Census Bureau website at http://www.census.gov/hhes/well-being/. Additional approaches to measuring well-being not mentioned here (such as through opinion polls, standard budgets, etc.) are discussed in working papers available on the Census Bureau website at http://www.census.gov/hhes/povmeas/publications/wp-other.html.
4.	This indicator (quoted verbatim) and the others bulleted above it are discussed more fully in Julie Siebens, Extended Measures of Well-Being: Living Conditions in the United States: 2011, U.S. Census Bureau, Household Economic Studies, P70-136, September 2013, http://www.census.gov/prod/2013pubs/p70-136.pdf.
5.	Health indicators, along with other indicators of well-being such as physical environment and safety, economic circumstances, family and social environment, behavior, and others, are included in the annual report, America's Children: Key National Indicators of Well-Being, available at http://www.childstats.gov/americaschildren/. Throughout the report, references are included for studies of the impacts of adverse circumstances quantified by the indicators.
6.	A discussion of this spatial mismatch—the lack of access to work, both in terms of distance and opportunity—is included in William Julius Wilson, When Work Disappears: The World of the New Urban Poor, Alfred A. Knopf, Inc., 1996.
7.	Mollie Orshansky, "Counting the Poor: Another Look at the Poverty Profile," Social Security Bulletin, vol. 28 no. 1, January 1965, p. 3.
8.	The treatment of assets in the context of family hardships, and the advantages and disadvantages of including them in a poverty measure, was discussed in further detail by a panel from the National Academy of Sciences. Constance Citro and Robert Michael, eds., Measuring Poverty: A New Approach, Panel on Poverty and Family Assistance: Concepts, Information Needs, and Measurement Methods, "A Crisis Definition of Resources," National Academy Press, 1995, pp. 214-218. The Supplemental Poverty Measure, however, does consider homeownership as being related to the cost of basic needs; this is discussed more fully later in this report, in "The Supplemental Poverty Measure."
9.	Statistical Policy Directive 14, issued by the Office of Management and Budget, referred to "the statistics on poverty contained in the Census Bureau's Current Population Reports, Series P-60, No. 68," which used the CPS ASEC (then called the March Supplement) as the data source, https://www.census.gov/hhes/povmeas/methodology/ombdir14.html. The methodological approach used in the official measure is also applied using other surveys in order to obtain estimates for geographic levels not available from the CPS ASEC, though differences in data collection methods across surveys typically result in slightly different poverty estimates. These differences are discussed further in the "Data Sources Used in Poverty Measurement" section of this report.
10.	The Census Bureau poverty thresholds, discussed in this report, are different from the administrative poverty guidelines published by the Department of Health and Human Services (HHS). The HHS poverty guidelines are a simplification of the poverty thresholds, are rounded, increase by an equal dollar amount for each additional family member, and contain separate figures for Alaska and Hawaii. The guidelines, and various multiples thereof, are used in determining eligibility for many of the government's aid programs, but are not used to count the number of people in poverty or perform analyses of the poverty population.
11.	Gordon Fisher, "The Development of the Orshansky Thresholds and Their Subsequent History as the Official U.S. Poverty Measure," 1992, rev. 1997, http://www.census.gov/hhes/povmeas/publications/orshansky.html.
12.	Consumer Food Economics Research Division, Agricultural Research Service, U.S. Department of Agriculture, "Family Food Plans, Revised 1964," Family Economics Review, 1964 no. 3 (October 1964), p. 12.
13.	Eloise Cofer, Eleanor Grossman, and Faith Clark, "Family Food Plans and Food Costs." Home Economics Research Report No. 20, Agricultural Research Service, U.S. Department of Agriculture, November 1962.
14.	The revisions were eliminating distinctions between farm and nonfarm families, eliminating distinctions between female-householder families and all other families, and extending the poverty threshold matrix so that the largest category became families with nine or more members (previously it had been seven or more members). Details on these revisions are available in the section "Changes in the Definition of Poverty" in the U.S. Census Bureau's poverty report for 1980. The overall effect on the count of persons in poverty was small—an increase of 368,000 nationally, raising the measured number of poor from 29.3 million to 29.6 million, or an increase of 0.2 percentage points (from 13.0% to 13.2% of the U.S. population). See Carol Fendler, Characteristics of the Population Below the Poverty Level: 1980. U.S. Census Bureau, Current Population Reports, P60-133, July 1982, pp. 2-5. http://www.census.gov/hhes/www/poverty/about/history/p60-133.pdf#page=9
15.	This report focuses on two measures—the official poverty measure and the Supplemental Poverty Measure—but poverty measurement research has been conducted for decades and dozens of alternative measures have been developed. This report discusses the Supplemental Poverty Measure instead of the other measures because it uses the most current methodology, has received formal vetting, and encapsulates the measurement issues that were researched in the previous decades.
16.	Some other surveys ask about each member's relationship to the householder. Those using this approach are easier to administer (particularly with those filled out by respondents themselves without the aid of trained representatives), but they provide somewhat less family relationship detail. While the CPS ASEC is the data source used for the official poverty estimates for the nation, other data sources are used for smaller geographic areas and tracking economic changes among individual respondents. The Census Bureau provides guidance about the use of poverty data sources at http://www.census.gov/topics/income-poverty/poverty/guidance/data-sources.html.
17.	The Supplemental Poverty Measure applies a method of defining family relationship that is different from the official measure. Details are discussed in the "The Supplemental Poverty Measure" section of this report.
18.	Criticisms have been discussed in the mainstream press as well as within academia. A 1988 article (Spencer Rich, "Drawing the Line Between Rich, Poor," The Washington Post, September 23, 1988, https://www.washingtonpost.com/archive/politics/1988/09/23/drawing-the-line-between-rich-poor/60f5dbeb-dab3-4a42-819a-2dea34e7854e/) documented dissatisfaction about the official measure. This came from both those claiming it was too high, citing its failure to capture the effects of in-kind benefits for the poor and its overstatement of inflation, and those claiming it was too low, based on the fact that if the thresholds were derived using more recent household consumption data, they would be based on roughly five times the cost of food, not three times as Orshansky had computed in the early 1960s.
19.	See, for instance, CRS Report R43805, The Earned Income Tax Credit (EITC): An Overview, by Gene Falk and Margot L. Crandall-Hollick.
20.	Three such examples, in different decades, illustrate the participation from academic researchers and federal agencies in the research discussion: In 1976, the Department of Health, Education, and Welfare issued a multi-volume report entitled The Measure of Poverty: A Report to Congress as Mandated by the Education Amendments of 1974 in order to comply with Section 823 of P.L. 93-380. Authors of the volumes of that report included analysts from the University of Michigan, private research institutions (Mathematica; Urban Systems Research and Engineering, Inc.), and federal agencies (Bureau of Labor Statistics; Census Bureau; Department of Agriculture; Department of Health, Education, and Welfare; Department of the Treasury; Social Security Administration). The report is reproduced on the Census Bureau website at http://www.census.gov/hhes/povmeas/methodology/interagency/index.html. In 1985, a conference was held to discuss methods of computing the value of noncash benefits (Conference on the Measurement of Noncash Benefits, December 12-14, 1985, Fort Magruder Inn and Conference Center, Williamsburg VA; proceedings and related technical papers available at http://www.census.gov/hhes/povmeas/methodology/williamsburg/index.html). Participants included "115 persons, including 23 from the Census Bureau.... 40 persons from universities and nonprofit research organizations, 16 persons from interest groups and other private sector organizations, and 36 persons from other government agencies and Congressional Committees" (conference proceedings, Preface, p. III). The third example is the National Academy of Sciences Panel on Poverty and Family Assistance: Concepts, Information Needs, and Measurement Methods, which met from 1992 to 1995 and published its report, Measuring Poverty: A New Approach, in 1995. That publication provided the overall framework that, after approximately a decade and a half of research and refinement, was used to create the Supplemental Poverty Measure.
21.	The Federal Register notice, the comments it received, and responses thereto, have been published on the Census Bureau's website at http://www.census.gov/hhes/povmeas/methodology/supplemental//spm_fedregister.html.
22.	For details, see Trudi Renwick and Liana Fox, The Supplemental Poverty Measure: 2015, U.S. Census Bureau, Current Population Reports, P60-258RV, September 2016, http://www.census.gov/content/dam/Census/library/publications/2015/demo/p60-258.pdf.
23.	For details, see Trudi Renwick and Liana Fox, The Supplemental Poverty Measure: 2015, U.S. Census Bureau, Current Population Reports, P60-258RV, September 2016, http://www.census.gov/content/dam/Census/library/publications/2015/demo/p60-258.pdf.
24.	The text of OMB Statistical Policy Directive 14, issued in May 1978, has been posted on the Census Bureau website at http://www.census.gov/hhes/povmeas/methodology/ombdir14.html.
25.	In 2009, the Measuring American Poverty Act was submitted in the House as H.R. 2909, and in the Senate as S. 1625. Neither was passed. Details on the bills and how the poverty definition compared with the Supplemental Poverty Measure (specified by the OMB Interagency Technical Working Group) are available in CRS Report R41187, Poverty Measurement in the United States: History, Current Practice, and Proposed Changes, by Thomas Gabe.
26.	The text of OMB Statistical Policy Directive 14, issued in May 1978, has been posted on the Census Bureau website at http://www.census.gov/hhes/povmeas/methodology/ombdir14.html.
27.	For instance, the Statistics of Income from the Internal Revenue Service provided (and still provide) information based on tax returns, but several important types of income are not taxable, such as Supplemental Security Income, Workers' Compensation, and child support payments received. Another source of income data, the Personal Income accounts from the Bureau of Economic Analysis (formerly the Office of Business Economics), is used in the computation of Gross Domestic Product, but could not be used to measure poverty because the dataset lacks the family demographic information needed to assign the appropriate thresholds.
28.	Information on the CPS sample design is available from Technical Paper 66, available at http://www.census.gov/prod/2006pubs/tp-66.pdf.
29.	A census tract is a geographic area smaller than a county that does not cross county boundaries, and usually has a population between 1,200 and 8,000 people (about 4,000 optimally). For details, see the Census Bureau's website for "Geographic Terms and Concepts – Census Tract," at http://www.census.gov/geo/reference/gtc/gtc_ct.html.
30.	Poverty estimates from the decennial census and the CPS were slightly different: for 1999, the poverty rate was 12.4% according to Census 2000, and 11.9% according to the CPS. The disparities arose from differences in collection method, the amount of income detail obtained, and the ways the two surveys identified family units. The decennial census used a paper questionnaire designed to be filled out by the respondent on his or her own, while the CPS was administered by trained interviewers. The CPS questionnaire was more detailed than the decennial census, and asked about more types of income that a respondent may not remember to include on his or her own. The CPS also obtained more detail about the family relationships among all members of the household, as opposed to just the relationship to the householder that was obtained by the decennial census. As a result, the CPS identified "unrelated subfamilies," composed of people related to each other but not the householder, whose resources may be greater than the poverty level when combined together but not as single individuals.
31.	For details, see CRS Report R44461, Allocation of Funds Under Title I-A of the Elementary and Secondary Education Act, by Rebecca R. Skinner and Leah Rosenstiel.
32.	For a discussion of why SAIPE data are used in Title I-A funding allocations, and other frequently asked questions about SAIPE, see http://www.census.gov/did/www/saipe/about/faq.html#q2.
33.	For details on SAHIE, see http://www.census.gov/did/www/sahie/about/index.html; for details about the Medicaid expansion under the Affordable Care Act, see CRS In Focus IF10399, Overview of the ACA Medicaid Expansion, by Alison Mitchell; for details about CHIP, see CRS Report R43627, State Children's Health Insurance Program: An Overview, by Evelyne P. Baumrucker and Alison Mitchell.
34.	One exception is the American Community Survey (ACS), which, although it asks respondents to provide information on the past 12 months, actually documents a 23-month period because the survey is administered each month. Respondents surveyed in January 2014 provide information from January 2013 to December 2013 (the past 12 months). Respondents surveyed in December 2014 provide data for December 2013 through November 2014. The ACS data published for 2014 reflect all respondents surveyed throughout 2014, and therefore capture the poverty status of respondents from January 2013 to November 2014.
35.	The 1992 SIPP Panel was conducted from February 1992 to May 1995, and the 1993 SIPP Panel was conducted from February 1993 to January 1996. Using the data provided by these households in the SIPP, plus the new interviews of these households in the SPD from 1997 to 2002, illustrates dynamics over a 10-year period.
36.	Other examples of need-based programs that use the HHS poverty guidelines are listed in Appendix D of CRS Report R43863, Federal Benefits and Services for People with Low Income: Programs and Spending, FY2008-FY2013, by Karen Spar and Gene Falk.
37.	See P.L. 97-35, §673(2), reauthorized P.L. 105-285, §201.
38.	The Poverty Guidelines page on the HHS website explains that "FPL" has been used loosely to refer to the guidelines, but also cautions against that phrase's ambiguity. See https://aspe.hhs.gov/poverty-guidelines.
39.	A synopsis of the early history of the guidelines appears in Israel Putnam, "Poverty Thresholds: Their History and Future Development," included in The Measure of Poverty, Technical Paper I, "Documentation of Background Information and Rationale for Current Poverty Matrix," Mollie Orshansky, ed., U.S. Department of Health, Education, and Welfare, 1976, pp. 272-283, http://www.census.gov/hhes/povmeas/methodology/interagency/index.html.
40.	The method for computing the HHS poverty guidelines—with the steps in the arithmetic shown for 2016—is available at https://aspe.hhs.gov/basic-report/computations-2016-poverty-guidelines.