Introduction

The primary source of federal aid to elementary and secondary education is the Elementary and Secondary Education Act (ESEA)—particularly its Title I-A program, which authorizes federal aid for the education of disadvantaged students. The ESEA was initially enacted in 1965 (P.L. 89-10) "to strengthen and improve educational quality and educational opportunities in the Nation's elementary and secondary schools." The Title I-A program in particular provides supplementary educational and related services to low-achieving and other students attending elementary and secondary schools with relatively high concentrations of students from low-income families, as well as eligible students who live in the areas served by these public schools but attend private schools.¹ Title I-A is also a vehicle to which a number of requirements affecting broad aspects of public elementary and secondary education for all students have been attached as conditions for receiving grants. The ESEA was most recently comprehensively amended and reauthorized by the Every Student Succeeds Act (ESSA; P.L. 114-95). Appropriations for the ESEA for FY2020 were $25.9 billion. Of this, $16.3 billion was appropriated for the Title I-A program.

Since the enactment of the ESEA, Title I-A grants always have been calculated based on one or more measures of a child's family financial situation (also referred to as poverty measures), with an emphasis on providing aid to schools serving concentrations of children from low-income families. Currently, four individual formulas are used to determine Title I-A grants to local educational agencies (LEAs). These formulas are based on a variety of factors, including data available from the Small Area Income and Poverty Estimates (SAIPE) program, which is administered by the U.S. Census Bureau. SAIPE includes estimates of the number of children ages 5-17 living in families in poverty. These children account for about 97% of all children included in the Title I-A formula calculations (commonly referred to as formula children). However, SAIPE data are limited in their coverage of LEAs as SAIPE does not include LEAs that do not have traditional geographic boundaries, such as charter schools that are their own LEAs. Thus, adjustments for these and other LEAs (e.g., newly formed LEAs not included in SAIPE for a given year) must be made by state educational agencies (SEAs) using other data sources.

Questions have been raised over the years about the measure of poverty used for SAIPE population estimates that are a foundational element of the Title I-A allocation formulas, and whether these data are the most appropriate to use for determining LEA grant amounts. A summary discussion of these issues may be found in Appendix B.

Among ESEA programs, Title I-A is unique in its requirement that funds be provided to public schools based on a statutorily prescribed methodology. While the focus of this methodology continues to be children from low-income families, SAIPE data generally are not available at the school-level, so LEAs are required by law to use other sources of a family's income status to distribute Title I-A funds to schools.² For decades, the primary source of data has generally been students eligible for free and reduced-price lunch (FRPL) under the National School Lunch Program (NSLP). While the use of this measure of the number of children from low-income families always has had caveats,³ it has become increasingly complicated due to the implementation of the Community Eligibility Provision (CEP) under the NSLP. Schools participating in CEP no longer identify children as being from low-income families in the same way as in past years. This has implications for identifying the number of children from low-income families in a given school for grant distribution purposes as well as for disaggregating data on student performance based on whether a student is from a low-income family. As a result, consideration is being given to whether there is a more effective and reliable way to measure low-income status at the school level, as well as questions about whether relying only on whether a child is from a low-income family is the most appropriate way to identify disadvantaged children for the purposes of the Title I-A program.

This report begins with an overview of the Title I-A formulas used by the U.S. Department of Education (ED) to make grants to LEAs, with a focus on the numerous ways in which measures of a child's family income are employed. This is followed by an historical overview of the Title I-A formulas and how the measures of poverty included in them have changed over time. The discussion then focuses on measures of school-level poverty that are used to make Title I-A grants from LEAs to the school level. This includes a focus on allocation methods for the past 20 years and complications that have arisen related to continuing to use these allocation methods to make grants to schools and for Title I-A accountability and reporting requirements. The last section of the report discusses some alternative school-level poverty measures that ED has considered or is currently examining. The report also includes two appendices. Appendix A provides an overview of the key factors included in each of the Title I-A formulas. Appendix B provides a detailed discussion about the standard federal poverty measure included in the SAIPE data used to determine Title I-A LEA grants.

Allocation of Title I-A Funds by ED to States and LEAs Under Current Law

This section provides an overview of the four formulas used to make Title I-A grants to LEAs and how measures of a child's family income are incorporated in multiple ways throughout these formulas. As discussed below, the number of children living in families in poverty is the primary income measure included in all four formulas and accounts for nearly all of the children included in the formula child count that is used across the formulas. All of the formulas include hold harmless provisions that are based on the percentage of formula children in a given LEA. Each formula also has minimum threshold criteria related to the formula child count or percentage of formula children that an LEA must meet to be considered eligible to receive a grant under that formula. Two of the four formulas apply additional weighting to the formula child count or percentage of formula children in an LEA to adjust grant amounts further.

The four formulas Title I-A uses for the allocation of funds to states and LEAs are the Basic Grant, Concentration Grant, Targeted Grant, and Education Finance Incentive Grant (EFIG) formulas. Individual LEAs may be eligible to receive grants under one, two, three, or all four of the formulas. Although portions of each year's appropriation are allocated separately under each of these formulas, when the funds reach LEAs they are combined and used jointly.

Under the current ESEA, appropriations for Basic Grants and Concentration Grants are limited to the amount provided for each grant in FY2001, while appropriations in excess of this amount are to be allocated under the Targeted Grant and EFIG formulas.⁴ In practice, under annual appropriations legislation for each year since FY2001, the amount allocated annually under the Concentration Grant formula has remained constant, the amount allocated as Basic Grants has declined somewhat (as across-the-board budget cuts for Title I-A overall in some years have been applied only to Basic Grants), and remaining funds have been equally split between the Targeted Grant and EFIG formulas. For the latest fiscal year (FY2020), 40% of Title I-A appropriations were allocated under the Basic Grant formula, 8% under the Concentration Grant formula, and 26% each under the Targeted Grant and EFIG formulas.

Common Formula Elements Related to a Child's Family Income

There are several elements included in all four of the Title I-A allocation formulas that are related to a child's family income. Each of these factors is discussed below. A table summarizing the Title I-A formula factors may be found in Appendix A.

Population Factor (also referred to as the formula child count): Each formula has a population factor, which is the same in all four formulas. Currently, this factor comprises children ages 5–17

• in poor families, as estimated annually by the SAIPE program and based on the Census Bureau's standard poverty income thresholds (these constitute 96.7% of all formula children for FY2020);
• in certain institutions for neglected or delinquent children and youth, or in certain foster homes (these constitute 3.3% of all formula children for FY2020); and
• in families receiving Temporary Assistance for Needy Families (TANF) payments with incomes above the poverty income level for a family of four (these constitute less than 0.1% of all formula children for FY2020).

The total number of children across these three groups will be referred to in this report as formula children. The report focuses primarily on the first category of Title I-A formula children—those ages 5-17 in poor families, as estimated under the SAIPE program. As indicated above, almost all children considered in the allocation of Title I-A grants are included in this category. The evolution of this formula population category over time, in terms of both the definition of low family income/poverty as well as sources for population estimates, are discussed in more detail below.

The second category of children above—those in certain institutions for neglected or delinquent children and youth or in certain foster homes—are included as a separate group because of their special needs and because they are generally not included in the estimated poverty counts, even though their income level may be low. They will not be discussed further in this report because they are a relatively small and comparatively constant portion of the Title I-A allocation formula population.

The third category of children above—those in families receiving TANF payments above the poverty level for a family of four—are discussed briefly in historical context. In any recent year they have constituted a small portion of the children considered in the allocation of Title I-A funds.

LEA Minimum Eligibility Threshold: Each formula has an eligibility threshold for LEAs, which is a minimum number formula children, or a minimum formula child rate (formula children as a percentage of total school-age children), that must be met to be eligible for grants in most cases. The LEA minimum eligibility threshold varies by formula: it is 10 formula children and a school-age child poverty rate of either 2% for Basic Grants, or 5% for the Targeted Grant and EFIG formulas. For Concentration Grants, the LEA eligibility threshold is 6,500 formula children or a 15% school-age child poverty rate. With the partial exception of Concentration Grants, if an LEA does not meet the eligibility threshold for a given year, the LEA hold-harmless provision (see below) does not apply.

LEA Hold Harmless: Each of the formulas has a hold-harmless provision—a minimum annual grant level for LEAs. The hold harmless is a percentage of the previous year's grant under each formula that ranges from 85%-95% based on an LEA's formula child rate. Thus, hold-harmless provisions preserve 85%-95% of the previous year's funding levels for individual LEAs regardless of changes in the students counted under the allocation formulas, so long as LEA minimum eligibility thresholds are met.⁵

There also are two additional factors that are used in each of the Title I-A formulas that are not related to a child's family income. They are discussed briefly below.

Expenditure Factor: Under each of the formulas, the population factor is multiplied by an expenditure factor, which is based on state average expenditures per pupil (AEPP) for public K–12 education, subject to minimum and maximum levels. For all except the EFIG formula, the minimum AEPP is 80% and the maximum is 120% of the national average. For the EFIG formula, the minimum is 85% and the maximum is 115% of the national average. These amounts are further multiplied by a federal share of 0.4 to determine maximum authorized grants.

State Minimum Grant: In general, no state is to receive less than approximately 0.25% of total allocated Title I-A funds in amounts up to the FY2001 appropriation level, and approximately 0.35% of funds above that level, applied separately to each formula.

Under all four formulas, amounts determined on the basis of the formula factors described above are reduced proportionally to the aggregate level of available funds, subject to LEA hold-harmless and state minimum grant provisions.

Distinctive Elements of the Targeted Grant and EFIG Formulas Related to a Child's Family Income

In addition to these common elements, the Targeted Grant and EFIG formulas include other factors focused on formula child counts or rates.

For all stages in the allocation of funds under the Targeted Grant formula, as well as the allocation of state total grants to LEAs under the EFIG formula, the formula children are assigned weights on the basis of each LEA's number of formula children or formula child rate. As a result, the higher an LEA's formula child number or rate is, the higher grants per child counted in the formula it will receive. Under the Targeted Grant formula, the weighting factors are applied in the same manner nationwide; formula children in LEAs with the highest formula child numbers have a weight of up to 3.0, and those in LEAs with the highest rates of such children have a weight of up to 4.0, compared to a weight of 1.0 for formula children in LEAs with the lowest formula child number and rate of such children. In contrast, under the EFIG formula the degree of targeting (in terms of the ratio of the highest to the lowest weight) varies depending on the value of each state's equity factor (described below). Under both formulas, the higher of the two weighted child counts (on the basis of numbers or rates) is used in calculating grants for each LEA.⁶

The EFIG equity factor is based on a measure of the average disparity in expenditures per pupil among each of the LEAs of a state and the state average per pupil expenditure.⁷ This measure is referred to as the coefficient of variation (CV). In the CV calculations for this formula, an extra weight (1.4 vs. 1.0) is applied to estimated counts of formula children. Thus, the CV for a state would be minimized if it spent, from state and local funds, exactly 40% more per formula child compared to spending on other students, on average. In calculating grants, the equity factor is subtracted from 1.30. Typical state equity factors range from 0.0 (for the single-LEA jurisdictions of Hawaii, Puerto Rico, and the District of Columbia, where by definition there is no variation among LEAs), to approximately 0.25 for a state with high levels of variation in expenditures per pupil among its LEAs; the equity factors for most states fall into the 0.10 to 0.20 range.⁸ Thus, the multiplier (1.30 minus equity factor) typically ranges from 1.05 to 1.30. As a result, the lower a state's weighted expenditure disparities among its LEAs are, the lower its CV and equity factor, the higher its multiplier, and the higher its state total grant will be. Conversely, the greater a state's weighted expenditure disparities among its LEAs are, the higher its CV and equity factor, the lower its multiplier, and the lower its state total grant will be.

The EFIG effort factor, while not related to a child's family income, is based on a comparison of state average expenditures per pupil for public K-12 education with state personal income per capita. This ratio for each state is further compared to the national average ratio, resulting in an index number that is greater than 1.0 for states where the ratio of expenditures per pupil for public K-12 education to personal income per capita is greater than average for the nation as a whole, and below 1.0 for states where the ratio is less than the average for the nation as a whole. Narrow bounds of 0.95 and 1.05 are placed on the resulting multiplier, so that its effect on state grants is limited.

The EFIG formula also differs from the other three formulas in terms of being a two-stage formula. Under the Basic Grant, Concentration Grant, and Targeted Grant formulas, maximum grants are calculated for LEAs by multiplying the population factor by the expenditure factor for all LEAs meeting the minimum eligibility thresholds. For the EFIG formula only, in the first stage, state total grants are calculated in proportion to each state's total population factor (unweighted) multiplied by its expenditure factor, by 1.3 minus its equity factor, and by its effort factor. In the second stage, these state total grants are allocated to LEAs on the basis of a modified version of the formula child weighting scheme of the Targeted Grant formula, with the degree of targeting (the ratio of the weight applied to formula children in the highest poverty ranges compared to the weight for such children in the lowest poverty ranges) varying in three stages. The stage, or degree of targeting, used for substate allocation varies depending on each state's equity factor: the higher the equity factor (and therefore the greater the disparities in expenditures per pupil among a state's LEAs) is, the greater the degree of targeting on high-poverty LEAs in the intrastate allocation of EFIG funds will be.

One Program, Four Formulas

One reason for using four different formulas to allocate shares of the funds for a single program is that the formulas have distinct allocation patterns, which are intended to provide varying portions of funds to states and localities with differing indicators of need for assistance. In addition, some of the formulas contain elements that are deemed to have desirable incentive effects or to be significant symbolically in addition to their impact on allocation patterns (e.g., the equity and effort factors in the EFIG formula). Finally, there is an explanation for the use of four different formulas based on legislative history: the Targeted Grant and EFIG formulas were initially proposed as replacements for the Basic Grant plus Concentration Grant formulas (i.e., the Targeted Grant and EFIG formulas were each originally intended to be the sole Title I-A formula). However, as proposals were debated and compromised in the development of the Improving America's Schools Act (P.L. 103-382), which reauthorized the ESEA in 1994, both of these formulas were ultimately established to complement, but not replace, the Basic Grant and Concentration Grant formulas.

SAIPE Data

As previously discussed, the primary factor used to determine an LEA's formula child count is the number of children ages 5-17 in an LEA living in families in poverty. This number is estimated annually by the Census Bureau based on the number of children ages 5-17 living in an LEA, regardless of whether they attend a public or private school, or no school at all, and whether or not the public school they attend is operated by a traditional, geographically based LEA, a regional LEA providing certain types of education (e.g., vocational-technical education) to students in multiple traditional LEAs, a school in a different LEA under a multiple-LEA or statewide choice program, or a charter school that is treated as a separate LEA under state law and enrolls students who reside within the boundaries of one or more traditional LEAs. In making these estimates, however, the Census Bureau is only able to include LEAs that have traditional geographic boundaries. If the Census Bureau tried to include other entities, such as charter schools, which are considered to be LEAs under the laws of many states, the same children could be counted multiple times. For example, for the District of Columbia, the SAIPE data includes an estimate of the number of children ages 5-17 living in families in poverty for the District of Columbia Public Schools (DCPS). The District of Columbia also includes dozens of charter schools that are considered their own LEAs. If the Census Bureau tried to estimate the number of such children living in the boundaries of each charter school, each child ages 5-17 living in families in the District of Columbia would be counted dozens of times as the geographic boundaries for the charter schools that are their own LEAs are the same as the geographic boundaries for DCPS.

Due to these limitations, regulations require state educational agencies (SEAs) to adjust the LEA allocations calculated by ED on the basis of traditional LEAs to shift shares of those grants to LEAs that are not included in ED's allocation procedures, including charter school LEAs.⁹ As SAIPE data are not available to make these determinations, SEAs must use other measures of a child's family income to determine the number of relevant children in each of the LEAs for which ED was unable to calculate grants. This topic is discussed further below.

Measures of Family Income Used as the Primary Population Factor in Allocation of Title I-A Funds to LEAs Since Enactment of the ESEA¹⁰

When the ESEA was initially adopted in 1965, there was no official or standard measure of poverty in use by the federal government. As is discussed further below, the poverty measure now used by the Census Bureau and other federal agencies began to be developed in the early 1960s, but was adopted as the official federal measure of poverty in 1969. Largely as a result of this, the initial versions of Title I-A relied on single, fixed thresholds of income to define school-age children in low-income families. Until adoption of the standard federal poverty measure for Title I-A allocations in the ESEA amendments of 1974,¹¹ the income thresholds used in Title I-A did not vary by family size, nor were they effectively updated at any time between 1965 and 1974. After adoption of the 1974 amendments to the ESEA, subsequent ESEA amendments made major changes to some of the measures of family income used in the Title I-A allocation formulas, as well as the geographic level at which the grants were calculated by ED and the frequency with which allocation population data were updated. The Title I-A allocation population factors required under the original ESEA and subsequent amendments are discussed below.

The Original ESEA of 1965

The original ESEA of 1965 (P.L. 89-10) established a low-income threshold of $2,000. Thus, counts of children ages 5-17 from low-income families used to calculate Title I-A grants were those in families with income below $2,000, according to the 1960 Census.

At the same time, it was understood that many families with school-age children had incomes above $2,000 per year due primarily or solely to financial assistance received under the income support program authorized by Title IV, Part A of the Social Security Act (SSA), then known as Aid to Families with Dependent Children (AFDC).¹² There was also interest in incorporating a population factor that would be updated more frequently than the Decennial Census; the AFDC counts were to be updated annually, based on data collected by what was then the Department of Health, Education, and Welfare (DHEW).¹³ Thus, it was decided that estimates of the number of school-age children in families with income below $2,000 would be supplemented by counts of such children in families receiving AFDC payments above $2,000.¹⁴

The initial ESEA (as well as every subsequent revision of it) provided that allocation formula data should be compiled, and grants be calculated by the federal government, on the basis of LEAs, if satisfactory LEA-level population data were available. In 1965, and for many years thereafter—until FY1999, as discussed below—such satisfactory LEA-level population data were not available, and grants were calculated by the U.S. Office of Education (until 1979)/Department of Education (subsequently) on a county basis, with sub-county grants to LEAs calculated and distributed by SEAs.

Reauthorizations of the ESEA: 1966-1988

Subsequent to the adoption of the ESEA in 1965, Title I-A was revised with respect to the allocation formula population requirements on numerous occasions. These revisions are briefly outlined below.

1966—Elementary and Secondary Education Amendments of 1966 (P.L. 89-750)

Counts of children ages 5-17 who are neglected, delinquent, and in foster care were first added to those from families with income below $2,000 and in families receiving AFDC payments above $2,000.¹⁵ This amendment also provided that the low-income threshold be raised from $2,000 to $3,000 for the allocation of funds appropriated for FY1968 and beyond, but this increase was never implemented.¹⁶

1967/1968—Elementary and Secondary Education Act Amendments of 1967 (P.L. 90-247)

The provision for increasing the low-income threshold from $2,000 to $3,000 was delayed until maximum authorized payments based on estimates of school-age children in families with income below the $2,000 level were provided, which never occurred.¹⁷

1969/1970—Elementary and Secondary Education Act Amendments of 1969 (P.L. 91-230)

These amendments provided for increases in the low-income threshold—to $3,000 for FY1972, and to $4,000 thereafter.¹⁸ Again, these increases would only take effect after maximum authorized payments based on the $2,000 level were provided, which did not occur.

1974—Education Amendments of 1974 (P.L. 93-380)

In the period leading up to consideration and adoption of the 1974 ESEA amendments, major shifts were projected to occur in the allocation of funds among regions, states, and counties nationwide. The primary cause of the changes in allocation patterns was replacement of estimates from the 1960 Census with estimates from the 1970 Census on children ages 5-17 in families with income below $2,000. The 1970 Census estimates of the number of such children were much lower, while the number of children in families receiving AFDC payments above $2,000, updated annually, continued to increase. As a result, the number of children counted under the AFDC factor was, for the first time, substantially higher than the number counted under the low-income factor.¹⁹ At the same time, a standard federal definition of poverty for statistical purposes had been adopted in 1969, as is discussed further below.²⁰

After extended congressional debate, much of it focused on estimates of the effects of possible allocation formula revisions on states and counties that were developed by the Congressional Research Service (CRS)²¹ and ED, the Title I-A population factor was changed to the total estimated number of children ages 5-17 in families with income below the standard federal poverty measure, specifically applying the poverty thresholds as used for the 1970 Census,²² plus the number of children ages 5-17 in families receiving AFDC payments above the standard federal poverty level for a non-farm family of four persons, multiplied by two-thirds.²³ A discussion of the development of the standard federal poverty measure, and issues related to it, is in Appendix B.

1978—Education Amendments of 1978 (P.L. 95-561)

The AFDC child count was changed from two-thirds of children ages 5-17 in families receiving AFDC payments with incomes above the standard federal poverty level for a non-farm family of four persons to 100% of such children.²⁴ The AFDC program was replaced by the TANF program in 1996.²⁵ This provision remains unchanged to the present. However, in years subsequent to the adoption of the 1978 ESEA amendments—as poverty income thresholds rose faster than AFDC, then TANF, payments—the number of children counted in the Title I-A formulas because their families receive AFDC/TANF payments yet have incomes above the poverty level for a family of four has steadily declined so as to become virtually nonexistent. For FY2020 grants, the number of such children constituted less than 0.1% of all children counted in the Title I-A allocation formulas.

The 1978 amendments provided that estimates on school-age children in poor families from the 1970 Census be replaced with more recent Census estimates when those became available.²⁶ In addition, the 1978 amendments provided that a portion of future appropriation increases would be allocated at the state level (only) using a different measure of low-income and a different source of population data. This change was made primarily in order to utilize population data more current than that available from the 1970 Census.²⁷ One-half of appropriation increases over the FY1979 level was to be allocated to states on the basis of children ages 5-17 in families with income below 50% of the (national) median income for four-person families. The source of these estimates was not to be the Decennial Census but rather a one-time Survey of Income and Education (SIE) conducted by the Census Bureau in 1976 (based on 1975 income).²⁸ These state total grants were then to be allocated within states in proportion to the remaining Basic Grants (based on Census and AFDC population data).²⁹

1988—Augustus F. Hawkins-Robert T. Stafford Elementary and Secondary School Improvement Amendments of 1988 (P.L. 100-297)

The provision for allocation of a portion of Title I-A grants to states on the basis of population estimates from the 1976 Survey of Income and Education was dropped. Also eliminated were references to 1970 poverty thresholds, so the most recent Census poverty income thresholds could be applied to 1980 and subsequent Census data.³⁰

The Improving America's Schools Act of 1994: Grants by LEA and Poverty Population Updates

The 1994 amendments to the ESEA—the Improving America's Schools Act (IASA; P.L. 103-382)—provided for major changes in the source of estimates for children ages 5-17 in poor families, and in the geographic level at which grants would be calculated by the federal government.

As discussed earlier, from the beginning of Title I-A in 1965, grants were calculated by the federal government on a county basis, with sub-county allocations to LEAs calculated by SEAs. From 1965 onward, the primary population factor data on school-age children in low-income/poor families for counties were updated only once every 10 years,³¹ when data from the Decennial Census became available.

No data from the 1960 Census were compiled by the Census Bureau by LEA. Data were compiled by LEA from the 1970 Census and 1980 Census, but these were not deemed to be sufficiently reliable to be used for the purpose of allocating Title I-A grants. However, in the 1990s, the Census Bureau initiated efforts both to compile selected population data for LEAs and to update these data more frequently than once per decade, through what became known as the SAIPE program.³² The provision for use of these population updates was added to Title I-A in an attempt to distribute funds on the basis of the latest available, reliable data on the distribution of school age children in poor families among states and localities, and to try to minimize the considerable disruption that had occurred previously with the introduction of new population data from the Decennial Census.

As amended by the IASA in 1994,³³ the Title I-A statute provided that beginning in FY1997, the Secretary of Education "shall" use updated population data prepared by the Census Bureau "unless the Secretary [of Education] and the Secretary of Commerce determine that use of the updated population data would be inappropriate or unreliable, taking into consideration the recommendations" of a series of studies of the updating methodology and process to be conducted by the National Academy of Sciences (NAS).³⁴ In March 1997, an NAS panel³⁵ recommended use of a combination of 1990 Census and income year (IY)1993 updated population estimates in allocating FY1997 (1997-1998) Title I-A grants.³⁶ Subsequently, the NAS panel recommended the use of a revised set of IY1993 SAIPE estimates as the sole basis for calculating FY1998 grants, and ED followed this recommendation as well.³⁷

These grants continued to be calculated by ED on a county level. However, beginning with FY1999 grants, the NAS panel recommended³⁸ that ED use the latest available SAIPE estimates of school-age children in poor families and that grants be calculated by ED on the basis of LEA, not county, population estimates from SAIPE, and ED has since followed these recommendations.

Use of SAIPE Data to Calculate Title I-A Grants to LEAs

Initially, SAIPE provided estimates of population data used to calculate Title I-A grants—total children ages 5-17, related children ages 5-17 in poor families, and total population (all ages)—every second year, starting with estimates for IY1993. Beginning with data for IY1999, however, SAIPE estimates have been prepared annually. As of the cover date of this report, the latest published SAIPE estimates are for IY2018 and were published in December 2019.³⁹

SAIPE estimates are available at the state, county, and LEA levels. SAIPE is not a survey of households separate from other federal surveys by the Census Bureau or other agencies.⁴⁰ Rather, SAIPE estimates are produced through statistical modeling and adjustment of administrative data along with survey data. The adjustments and modeling are intended to produce estimates for small areas such as LEAs that are more reliable and accurate than estimates based only on survey data. The administrative data used for SAIPE include tax returns from the Internal Revenue Service (IRS), and participation data for the Supplemental Nutrition Assistance Program (SNAP)⁴¹ and the Supplemental Security Income (SSI) program.⁴² The survey data include personal income data from the Bureau of Economic Analysis, Decennial Census and annual population estimates, and data from the American Community Survey (ACS).⁴³ LEA boundaries are provided by the Census Bureau's School District Review Program (SDRP).⁴⁴

More specifically, SAIPE estimates for states and counties are based on ACS population samples, supplemented by administrative data (IRS data and SNAP participation data). SAIPE estimates for LEAs are based on ACS population samples, IRS data, and population and poverty estimates for the counties in which LEAs are located. The total number of children ages 5-17, and the number of such children in poor families, are estimated for counties using a weighted combination of regression modelling (based on administrative data from IRS and SNAP) and direct estimates from sample surveys. LEA estimates are based on shares of the populations within counties.⁴⁵

Subsequent Reauthorizations of the ESEA

Subsequent reauthorizations of the ESEA by the No Child Left Behind Act (NCLB; P.L. 107-110) and the Every Student Succeeds Act (ESSA; P.L. 114-95) did not make changes to the poverty measures included in the Title I-A formulas.

ESEA Title I-A Grants to Schools

Unlike other federal elementary and secondary education programs, under which grants are made to LEAs or states, most Title I-A funds have always been allocated to individual schools, although LEAs retain substantial discretion to control the use of a share of Title I-A grants at a central district level. In almost all cases, the data used to determine which students are from low-income families for the selection of participating schools and distribution of funds among them are not the same as those used to estimate the number of children ages 5-17 living in families in poverty for purposes of calculating allocations to states and LEAs. This is because Census data are generally not available on the number of school-age children enrolled in a school, or living in a residential school attendance zone, with income below the standard federal poverty threshold. Thus, LEAs use available proxies for low-income status.

History of Title I-A Provisions for School Selection, Identification of Low-Income Students within Schools, and Allocation of Funds

The original version of the ESEA, enacted in 1965, contained only broad and unspecific provisions regarding the selection of schools to participate in Title I-A. The statute provided that "payments under this part will be used for programs and projects ... which are designed to meet the special educational needs of educationally deprived children in school attendance areas having high concentrations of children from low-income families."⁴⁶

This statutory language was supplemented by policy guidance from the U.S. Office of Education (USOE). By the mid-1970s, the USOE guidance⁴⁷ provided that LEAs could, with approval by their SEA, use any of the following measures of low family income (or a combination of such measures) to select schools to participate in Title I-A:

• number of children in families receiving AFDC payments,
• number of children in families with income below the poverty level as used by the Census Bureau,
• number of children eligible for free or reduced-price school lunch;
• number of children in low-income families according to school surveys,
• "health statistics,"
• "housing statistics,"
• "employment statistics," or
• "other" data sources.⁴⁸

Whatever measure was chosen, the same measure had to be applied to all public schools in the LEA, although LEAs could choose to focus Title I-A services on only one or more grade levels (e.g., only elementary schools). In addition, schools could be selected for participation on the basis of either the percentage or number of students from low-income families either residing in school attendance areas or actually enrolled in schools. There was also a limited option to serve schools with higher incidences of educationally deprived students (a term not specifically defined) over schools with higher concentrations of students from low-income families. The provisions for allocation of Title I-A funds among eligible schools were also broad—funds were to be allocated in proportion to the number of students to be served and their educational needs, as determined by the LEA. A 1977 report by the National Institute of Education found that "the number of free lunch recipients is the most readily available source of poverty data for most school districts," and that 66% of a sample of LEAs used these data for school selection and allocations, while 51% of LEAs used AFDC data, and some LEAs were found to use a combination of free lunch, AFDC, and other authorized data sources.⁴⁹

This pattern of broad authority for LEAs to select measures of low family income for school selection and funding allocation under Title I-A continued until the ESEA was reauthorized by the IASA in 1994. The IASA contained the first explicit statutory specification of the data that may be used for selection of schools to participate in Title I-A and allocation of funds among them.⁵⁰ It provided that schools are to be selected on the basis of their percentage (not number) of children from low-income families, while funds are to be allocated among eligible schools on the basis of their number children from low-income families.⁵¹ With updates of programmatic references, the IASA provisions remain in effect today, and are described below.

Current Provisions for Selection of Participating Schools and Allocation of Funds Among Them

The general policy for Title I-A school selection and allocation established under the IASA continues under the ESEA as most recently reauthorized by the ESSA and related policy guidance.⁵² These provisions are described in greater detail below.

While there are several rules related to selection of schools to participate in Title I-A, LEAs must generally rank their public schools by their percentage of students from low-income families, and serve them in rank order. ⁵³ LEAs may choose to consider only schools serving selected grade levels (e.g., only elementary schools) in determining eligibility for grants, so long as all public schools with 75% or more of students from low-income families and, at the LEA's discretion, high schools with 50% or more students from low-income families, receive grants first, to the extent that funds are available.⁵⁴

All participating schools must generally have a percentage of children from low-income families that is equal to or above the LEA's average, or 35%, whichever is lower.⁵⁵ The percentage of students from low-income families for each public school is usually measured directly, although LEAs may choose to measure this percentage indirectly for middle or high schools, based on the measured percentages for the elementary or middle schools that students attended previously (feeder schools). LEAs have the option of setting school eligibility thresholds higher than the minimum in order to concentrate available funds on a smaller number of schools, and this is the practice especially in many large, urban LEAs.

The current Title I-A statute allows LEAs to use the following low-income measures for school selection and allocations:

• eligibility for FRPL under the federal child nutrition programs (the Richard B. Russell National School Lunch Act),
• eligibility for TANF,
• eligibility for Medicaid,
• Census poverty estimates (in the very rare instances where such estimates are available for individual schools or school attendance areas), or
• a composite of two or more of these measures.

While FRPL data are not used by ED to determine grants to LEAs, in school year 1997-1998, approximately 90% of LEAs receiving Title I-A funds were using this data—sometimes alone, sometimes in combination with other authorized criteria—to select Title I-A schools and allocate funds among them.⁵⁶ According to a more recent ED report, "school districts suballocate most of their Title I funds to eligible schools based on each school's number of low-income children, typically using data from the free or reduced-price lunch program."⁵⁷

Individual children become eligible for FRPL in one of two ways: (1) submission of application forms indicating that household income is below specified thresholds for free or reduced-price lunch eligibility; or (2) Direct Certification, through which eligibility for FRPL is established through household participation in one of a number of public benefit programs (see below). In addition, children may obtain access to free school meals through school or LEA participation in CEP (discussed later in this report). Overall, the national average percentage of public school students who are eligible to receive free or reduced price lunches was 52.3% for the 2016-2017 school year.⁵⁸

The income eligibility thresholds for households submitting applications specifically for free and reduced-price lunches are higher than the poverty levels used in the allocation formulas to states and LEAs: 130% of the poverty income threshold for free lunches, and 185% for reduced-price lunches.⁵⁹ From the perspective of consistency, it would be preferable to use the same measure of poverty in determining children to be counted in the allocation of Title I-A funds both to LEAs, and to individual schools within LEAs. However, that has thus far proven to be a practical impossibility. SAIPE and other Census data that provide estimates of school-age children in poor families are not designed to provide reliable estimates at levels of geography smaller than LEAs. Thus, it is currently possible only rare instances to obtain Census poverty estimates for individual school attendance areas. Until recently (as is discussed further below), data on the number of students eligible for or receiving free and/or reduced-price lunches has been the only indicator of student low-income status for individual schools, and those data are only available at the income thresholds of 130% and 185% of poverty, not 100%.

Data on students whose family income is at 100% of the poverty level could be collected for individual schools through state or local surveys of student family income, but such surveys are currently conducted by only a limited number of states and LEAs. Finally, one alternative to free and/or reduced-price lunch data currently under consideration in some LEAs is counts of children directly certified for free lunches through family participation in programs such as SNAP, among others. Eligibility for SNAP and other direct certification programs is determined by low family income, with somewhat varying income thresholds, plus a variety of other factors.

Focusing on Title I-A grants to LEAs, and to schools within LEAs, for the 2020-2021 school year, the grants to LEAs are based on SAIPE estimates published in December 2019, based on income in calendar year 2018. For that year, the weighted average poverty threshold (100%) for a four-person family was $25,701.⁶⁰ For that same year, in LEAs that use FRPL data to determine school-level allocations, grants to schools generally would be based on FRPL participation in the preceding school year, 2019-2020.⁶¹ For 2019-2020, the income eligibility thresholds for four-person families were $33,475 for free lunches, and $47,638 for reduced-price lunches.⁶²

As noted above, students may also qualify for free school meals, through direct certification, if their household participates in one of more of the following federal benefit programs: SNAP, TANF, Food Distribution Program on Indian Reservations, or Medicaid in areas approved for the U.S. Department of Agriculture's (USDA's) Medicaid Direct Certification Demonstration Projects.⁶³ Students may be approved as well without an application if they are in foster care or are migrant, homeless, runaways, or attending a Head Start program. Direct certification takes place through matching of school enrollment data with participation data for the indicated benefit programs. Income and status eligibility standards for these programs vary across different programs, and in some cases across different states.⁶⁴

While FRPL data have long been the primary basis for allocating Title I-A funds among schools—and for a variety of other purposes related to the identification of students from low-income families and of schools with large proportions of such students⁶⁵—this data source has always had a number of imperfections. For example, a 2015 USDA study, based on data from the 2012-2013 school year, concluded that an estimated 20.2% of FRPL applicants were placed in the wrong category (among the three categories of free, reduced-price, or denied). The authors of the USDA study further found that 60% of these errors were due to incorrect reporting of income or other information by applicant families, 30% were due to administrative errors, and 10% were due to a combination of these or miscellaneous causes. Administrative errors in the certification process were found to be much higher when participation was based on family paper applications submitted specifically for the school nutrition programs (an error rate of 14% of all applicants) than when participation was based on direct certification (through family participation in TANF or SNAP, for example), or under CEP (error rates of 4% and 2%, respectively).⁶⁶ Other studies have found that substantial proportions of students eligible to participate in FRPL do not do so, and/or their families do not apply for the program, and that the non-participation rates vary among regions of the nation, urban-suburban-rural locales, immigration/English Learner status, and age/grade levels of students.⁶⁷

Title I-A funds are allocated among participating schools in proportion to their number of students from low-income families, although grants to eligible schools per pupil from a low-income family need not be equal for all schools. LEAs may choose to provide higher grants per child from a low-income family to schools with higher percentages of such students. For example, an LEA could choose to provide higher grants per child to a school where 75% of students are from low-income families than to a school where 45% of students are from such families.

Determination of the Share of Title I-A Grants to Be Used to Serve Eligible Students Attending Private Schools

The share of funds to be used by each recipient LEA to serve educationally disadvantaged students attending private schools is determined based on the number of private school students from low-income families living in the residential areas served by public schools selected to receive Title I-A grants. LEAs may use for this purpose either the same source of data used to select and allocate funds among public schools (i.e., usually FRPL data) or one of a specified range of alternatives.⁶⁸ In cases where a state or LEA deems itself to be unable to provide Title I-A services to eligible private school students, or where the U.S. Secretary of Education determines that such services have been inadequate, the Secretary arranges for services to be provided via a bypass arrangement, under which the services are provided by a third-party entity.

Suballocation of Title I-A LEA Grants to Charter Schools and Other "Special LEAs"

Under current law, the allocation calculations by ED do not take into account charter schools that are treated under state law as separate LEAs, nor do they take into account LEAs that provide specialized services (such as vocational-technical education) to multiple traditional LEAs as no SAIPE data are available to calculate grants to these entities. Thus, per regulations, the grants as calculated by ED must be adjusted to provide funds to eligible LEAs in these categories, all of which are referred to in ED policy guidance as "special LEAs."⁶⁹ With respect to charter schools, these adjustments apply only to charter schools that are treated under state law as separate LEAs; charter schools that are not treated as separate LEAs under state law receive Title I-A grants in the same manner as other public schools within a traditional LEA.

ED's policy guidance⁷⁰ describes two different methods for determining Title I-A grants to charter school and other special LEAs, one for states that are able to determine the sending LEAs in which charter school and other special LEA students reside, and one to be used by states that do not have this information. Under both of these methods, SEAs must estimate the number of Census poverty children who enroll in a charter school LEA or other special LEA.⁷¹ As with the process of suballocating grants to schools within traditional LEAs, this is most often done with FRPL counts or other measures of low-income authorized for school-level allocations (discussed above).

Under the first method, each charter school or other special LEA reports to the SEA its total enrollment as well as its enrollment of students from low-income families, and identifies the traditional (geographically based) LEA in which each of these students resides. SEAs then use the ratio of FRPL (or other authorized measures of low-income) students to Census poverty children in the specific LEA in which each charter school student from a low-income family resides to estimate the number of Census poverty children (as counted when determining the federal allocation to the state) for each charter school or other special LEA. SEAs add to this Census poverty estimate for the charter school LEA the number of other formula children included in the Title I-A formulas⁷² to derive a total formula child count for each charter school LEA. For each such formula child, the charter school LEA receives an amount equal to the Title I-A grant per formula child associated with the sending LEA in which the child's family resides. At the same time, an equivalent amount is deducted from the grant for the sending LEA.

Under the second method, the enrollment data reported by charter schools and other special LEAs again are used to estimate the number of formula children for each special LEA, but in this case using the statewide average ratio of Census poverty and other formula children to FRPL (or other low-income measure) students. These formula child counts for each special LEA⁷³ are summed to determine the share of all formula children in the state who attend charter school and other special LEAs. In this case, the grants to all traditional LEAs in the state, not just the specific LEAs in which charter school students reside, are reduced by this percentage, and each special LEA receives a grant based on the statewide average Title I-A grant per formula child.

Recent Developments Regarding Data on Students Eligible for Free and Reduced-Price Lunches: The Community Eligibility Provision (CEP)⁷⁴

The Healthy, Hunger-Free Kids Act of 2010 (P.L. 111-296) created CEP as a new option for how schools can operate the National School Lunch and School Breakfast Programs authorized under the Richard B. Russell National School Lunch Act. Under CEP, free meals are provided to all students at participating schools.⁷⁵ CEP is designed to ensure access to school meals by students from low-income families, especially in high-poverty schools, and simplify administration of the school meal programs by eliminating the use of applications to collect family income information and the need to track children by eligibility category in the lunchroom.

Community eligibility was initially phased in in a few states at a time and became available in all states beginning with the 2014-2015 school year, when 14,214 schools serving more than 6.6 million children participated in the program. Since then, participation in CEP has increased steadily, rising to 28,614 schools, serving more than 13.6 million students in 4,698 LEAs in the 2018-2019 school year.⁷⁶ In that year, participating schools constituted an estimated 64.6% of all schools eligible to participate in CEP,⁷⁷ or 28.7% of all public schools.

Implementation of CEP by LEAs and schools has important implications for Title I-A. As discussed above, for the vast majority of public schools eligibility to receive FRPL has been the sole, or at least the primary, indicator of low family income under Title I-A. This data source is no longer available for 28.7% of all public schools as of the 2018-2019 school year, and that percentage is likely to continue to rise in the future. As is discussed below, alternatives to FRPL are available as indicators of the percentage of students from low-income families in CEP schools. However, on average these sources tend to include a more limited segment of the low-income population than FRPL, due to differences in income thresholds and other eligibility criteria, as well as differing program administrative procedures. Therefore, these alternative sources of data on students from low-income families in individual schools are not directly comparable to the FRPL counts used in the past for CEP schools, or used currently by most non-CEP schools. Such changes in the sources of data on students from low-income families, and use of different sources for different schools in the same LEA, can result in substantial shifts in the patterns of Title I-A eligibility and allocation levels among schools, and potentially inconsistent treatment of schools within the same LEA. Further, as is discussed later in this report, changes in the sources of data on students from low-income families affect policies requiring schools to be accountable for Title I-A grants by reporting achievement levels specifically for students from low-income families and taking appropriate actions when those achievement levels are inadequate

School meals data never have any effect on state total grants. In the great majority of cases, there is also no consideration of school meals data in the calculation of grants to LEAs. This section provides an overview of CEP, followed by discussion of the implications of this option for Title I-A.

Overview of CEP

A school, group of schools, or an entire LEA may operate under CEP if the LEA chooses to do so and if at least 40% of the total enrollment is approved for free school meals without an application. Students enrolled without an application are referred to as Identified Students because they have been identified by another program as low-income or especially vulnerable. The share of enrolled students who are Identified Students is referred to as the Identified Student Percentage (ISP). These counts of Identified Students are the alternative to FRPL counts (referred to above) that are now available for CEP schools, and potentially available for non-CEP schools as well.

Identified Students include those whose families receive SNAP benefits, TANF cash assistance, Food Distribution Program on Indian Reservations benefits, or Medicaid in areas approved for the USDA's Medicaid Direct Certification Demonstration Projects.⁷⁸ Such students are deemed to be directly certified for free lunches through family participation in the specified programs. Students may also be approved without an application if they are in foster care or are migrants, homeless, runaways, or attending a Head Start program. For purposes of CEP, ISPs need to be updated once every four years, although LEAs are encouraged to update these data more often.⁷⁹

It should be noted that SAIPE estimates of the number of school-age children in poor families, which are used to allocate Title I-A grants to states and LEAs, are more comprehensive than counts of Identified Students in schools under the CEP program. SAIPE estimates include all children ages 5-17 in families with income below the standard federal poverty level, regardless of whether they attend public or private schools, or any school at all; the immigration or citizenship status of the children or their parents; whether their parents have applied for assistance under any governmental aid program; the employment or educational status of their parents; and whether their parents may have been convicted of violating any laws. In contrast, parents must meet a number of eligibility criteria related to the above factors in order for the parents and children to receive assistance under SNAP or other programs under which students may be directly certified and thereby included in their schools' ISP. (For a discussion of eligibility criteria for SNAP, see CRS Report R42505, Supplemental Nutrition Assistance Program (SNAP): A Primer on Eligibility and Benefits, and CRS Report R46371, Serving Free School Meals through the Community Eligibility Provision (CEP): Background and Participation.) Similarly, data on students receiving free or reduced-price lunches are limited by either the requirement for parents to complete applications or to meet the eligibility standards for SNAP or other programs that are the basis for direct certification of eligibility for free lunches.

CEP schools serve meals free to all students and are reimbursed by USDA at the free meal rate based on the ISP multiplied by 1.6, which is intended to reflect the average ratio of the number of students receiving free or reduced-price meals to the number of Identified Students.⁸⁰ As a result, schools participating in CEP are fully reimbursed at the free meal rate for providing free school meals to all students if their ISP is 62.5% or above; if the ISP is between 40% and 62.5%, they must provide at least some revenue for school meals from other sources.⁸¹

CEP and Title I-A Implications

The implementation of CEP has implications with respect to the determination of Title I-A grants to the school level and accountability measures. First, CEP schools need to identify an alternative data source (other than FRPL eligibility) to determine school-level Title I-A grants. The data source selected, however, must allow for comparability between CEP and non-CEP schools to ensure fairness. Second, these data choices affect allocation of Title I-A funds among schools, and Title I-A policies requiring schools to be accountable for these funds by reporting achievement levels specifically for students from low-income families and taking appropriate actions when those achievement levels are inadequate, as discussed below.

U.S. Department of Education Policy Guidance on CEP

ED published detailed policy guidance on the administration of Title I-A in LEAs with one or more schools participating in CEP in March 2015.⁸² For schools participating in CEP, the counts of students approved to receive free and reduced-price meals discussed above are no longer available, though such schools do have counts of Identified Students.⁸³ ED's March 2015 policy guidance lists several alternative sources of data that states and LEAs may use for selection of Title I-A schools and allocating funds among them. These alternative data sources are listed below.

• Identified Student Percentage multiplied by 1.6. The 1.6 multiplier is an estimate of the ratio of the total number of students approved for FRPL to the number of students approved for free meals without an application (through direct certification). For schools that participate in CEP individually, this percentage is identical to the percentage of meals for which they can claim reimbursement at the free rate from USDA. A school that participates in CEP as part of a group must calculate this percentage for the individual school and uses this individual percentage in the context of Title I-A, while using its group percentage in the context of operating the school meal programs. Where the ISP multiplied by 1.6 is used for CEP schools, non-CEP schools in the LEA may use any other authorized data source, such as children approved to receive FRPL. Alternatively, LEAs may use the number of Identified Students multiplied by 1.6 as the share of the school's enrollment that are from low-income families for all schools, whether or not they participate in CEP, providing consistent treatment of all schools.
  
  This approach allows LEAs to continue using FRPL or other data they have used in the past for non-CEP schools, while using a measure for CEP schools that is, on average, comparable. In addition, LEAs could use the same measure—ISP multiplied by 1.6—for both CEP and non-CEP schools, ensuring comparability between them. One potential difficulty is that the estimated ratio of students approved for free meals without an application to FRPL students for the nation as a whole may not apply to individual schools or LEAs. Further, if LEAs use the ISP multiplied by 1.6 for all of their public schools, such use of a new data source may result in changes in the identification of Title I-A schools and allocation of funds among them.⁸⁴
• Identified Student Percentage without use of the 1.6 multiplier (if used consistently for both CEP and non-CEP schools). Because students approved for free meals without an application are, on average, a subset of students who would qualify for free or reduced-price school meals if their families completed an application, this approach to identifying low-income students will generally lower the percentage of students considered low income at all schools. Therefore, LEAs that adopt this approach may wish to adjust by funding schools whose shares of low-income students are lowered as a result. Under the approach, the same data are used for CEP and non-CEP schools, ensuring comparable treatment. However, this would generally require the use of new data sources for all schools, possibly resulting in shifts in the identification of Title I-A schools and allocation of funds among them.
• Shares of students from low-income families as determined by state or local income surveys. While ED's policy guidance discourages their use due to concerns about administrative burdens,⁸⁵ states and LEAs could design and administer income surveys to meet their needs, not only for Title I-A, but potentially for other programs (such as state school finance formulas) as well. Aside from administrative burdens, there may be issues with the reliability and accuracy of state or LEA income surveys.
• The Medicaid, TANF, Census (where available), or composite data authorized under the ESEA statute. These data sources are already explicitly authorized under the Title I-A statute—and have been for many years, though few LEAs have chosen to rely on them. Census data, in particular, are rarely available for individual schools. Note that data on the number of children whose families participate in SNAP are not included here, presumably because that program is not among those specified in the statutory text of the ESEA.

With some of the options allowed under ED's policy guidance for Title I-A school selection and allocations in LEAs with CEP and non-CEP schools, there is potential concern about a lack of comparability in the low-income student data used for CEP and non-CEP schools. LEAs following ED's CEP policy guidance will in some cases be using different data sources on low-income students for CEP and non-CEP schools (e.g., Identified Students * 1.6 for CEP schools and students approved for FRPL for non-CEP schools). Also, while the data for non-CEP schools are updated annually, the data for CEP schools might be updated only once every four years (although the ED guidance encourages more frequent updates).

In addition to public schools, private schools may be eligible to participate in CEP if their ISPs are 40% or more. Thus, in determining the share of their Title I-A grant that must be used to serve private school students, LEAs may have to consider a variety of scenarios in which public and private schools may or may not be CEP schools. Options for counting students in low-income families for CEP private schools include those described above for CEP public schools. Nevertheless, the basic principles regarding equitable determination of low-income student counts for participating public and private schools remain unchanged, and ED's March 2015 policy guidance offers several illustrative examples.

State Adjustments of LEA Grants as Calculated by ED

As noted previously, school meals data never have any effect on state total grants. In the great majority of cases, there is also no consideration of school meals data in the calculation of grants to LEAs. However, states may, with ED approval, reallocate grants among their small LEAs—defined as those serving areas with a total population of fewer than 20,000 persons. In states exercising this option, total funds calculated by ED for small LEAs are aggregated and then reallocated based on the statutory formulas, but using alternatives (approved by ED) to the statutory population factor. Currently, nine states (Alaska, Iowa, Kansas, Maine, Montana, Nebraska, New Hampshire, North Dakota, and Oklahoma)⁸⁶ exercise this option to use alternative population data for reallocation of funds among their small LEAs, and many of these states use counts of children eligible for FRPL as at least a partial population factor.

For states that use alternatives to Census poverty estimates to allocate Title I-A funds among their small LEAs, ED's March 2015 policy guidance provides two options in cases where those states use school meals data in their formulas. First, states may use the number of Identified Students as their measure of low family income for all schools in compiling data for affected LEAs. Second, states may use the number of Identified Students multiplied by 1.6 for CEP schools, and use either the number of Identified Students multiplied by 1.6 or an unduplicated combination of Identified Students plus FRPL students for non-CEP schools.

With respect to SEAs' possible use of FRPL data to adjust ED allocations for some of their LEAs due to recent boundary changes or creation of new LEAs, or for charter schools treated as separate LEAs under state law, ED's policy guidance would allow use of any of the alternatives to FRPL data for CEP schools discussed above.

Possible Impact on CEP of Recent Changes in the Supplemental Nutrition Assistance Program (SNAP)

SNAP, formerly known as the Food Stamp program, assists low-income households obtain a nutritionally adequate diet. As is discussed below, counts of children ages 5-17 in households participating in the SNAP program, and changes in those counts over time, affect the allocation of Title I-A funds to states and LEAs, as well as to individual schools.

SNAP policies assume that households devote 30% of their monthly cash income to the purchase of food.⁸⁷ If, given the household's income level, this contribution is insufficient to meet the full cost of a diet set at the level of USDA's Thrifty Food Plan, SNAP benefits are available to eligible households to help make up the difference.⁸⁸ SNAP benefits are provided via Electronic Benefit Transfer (EBT) cards, which operate similar to debit cards when used to purchase eligible foods at participating food retailers.

In general, to be eligible for SNAP benefits applicants must have gross income below 130% of the poverty level and/or net income (after taking into account a number of allowable deductions) below 100% of the poverty level, plus have limited financial resources (such as cash or money in a bank account).⁸⁹ Households may also be categorically eligible for SNAP if they receive benefits under SSI, TANF, or state-run General Assistance (GA) programs.⁹⁰

There are a number of additional eligibility requirements for SNAP beneficiaries. For example, only U.S. citizens and certain legally present noncitizens may receive benefits. In addition, a number of work/training requirements apply to all SNAP beneficiaries aged 16-59, and a special, additional set of work/training requirements apply specifically to Able Bodied Adults Without Dependents (ABAWD) aged 18-49. In particular, ABAWD will generally need to meet specific work or training requirements in order to receive SNAP benefits for more than three months in any three-year period. Persons may be excused from either the general requirements or the specific ABAWD requirements under specified conditions.⁹¹

SNAP programs are administered by state SNAP agencies,⁹² and policies regarding some aspects of eligibility and benefit amounts may vary somewhat among the states. One example is state implementation of categorical eligibility of households for SNAP if they already participate in certain other programs, including SSI, state-run GA programs, or TANF. With respect to TANF, states have been allowed, if they so choose, to confer SNAP eligibility on households based not only on cash assistance but also on receipt of a wider range of often low-cost TANF benefits or services, such as provision of a brochure or pamphlet, or referral to a telephone hotline for additional information.⁹³ These categorically eligible households are able to bypass the usual SNAP asset limits, although their net income still needs to be below the standard threshold for SNAP eligibility.⁹⁴

A number of changes to SNAP eligibility provisions have been proposed or adopted in 2019 or 2020. (A comprehensive or detailed description of changes, whether proposed or enacted, is beyond the scope of this report. They are discussed in CRS Report R42054, The Supplemental Nutrition Assistance Program (SNAP): Categorical Eligibility, and CRS Insight IN11250, USDA Domestic Food Assistance Programs' Response to COVID-19: P.L. 116-127, P.L. 116-136, and Related Efforts) Data on SNAP beneficiaries are incorporated into SAIPE estimates of children ages 5-17 in poor families for states, counties, and LEAs. In addition, counts of Identified Students via household participation in SNAP are a key element in determining school eligibility for the CEP child nutrition program. Even in schools that do not participate in CEP, students may become eligible for free school lunches through household participation in SNAP (direct certification). Thus, any change in the number of households with school-age children in the SNAP program could affect the allocation of Title I-A funds to states and LEAs, as well as to individual schools.

Title I-A Accountability, Reporting Requirements, and Data on Low-Income Students

To receive Title I-A funds, states, LEAs, and schools must comply with a number of educational accountability requirements relating to standards and assessments. Under the ESEA, states participating in Title I-A are required to develop and adopt standards and assessments in mathematics, reading, and science.⁹⁵ States are also required to establish long-term and interim goals, including goals related to performance on assessments. States must annually measure the performance of all students and each subgroup of students in schools relative to these goals using a set of state-developed indicators. Subgroups for accountability and reporting purposes include economically disadvantaged students, students from major racial and ethnic groups, students with disabilities, and students with limited English proficiency.⁹⁶ Thus, for outcome accountability and reporting purposes, schools, LEAs, and states must be able to identify economically disadvantaged students.

A state must use a set of indicators that are based, in part, on the long-term goals it established to measure annually the performance of all students and each subgroup of students to evaluate public schools.⁹⁷ These indicators must include the following:

• Student Proficiency on Reading/Language Arts (RLA) and Mathematics Assessments. For all public schools, student performance on the RLA and mathematics assessments as measured by student proficiency, and for high schools, this may also include a measure of student growth on such assessments;⁹⁸
• Measures of Student Growth or Another Indicator of School Performance. For public elementary and secondary schools that are not high schools, a measure of student growth or another indicator that allows for meaningful differentiation in school performance;⁹⁹
• Graduation Rates. For public high schools only;¹⁰⁰
• English Language Proficiency. For all public schools, English Learners' (ELs') progress in achieving English language proficiency;¹⁰¹ and
• School Quality or Student Success. For all public schools, at least one indicator of school quality or student success (e.g., a measure of student engagement, postsecondary readiness, school climate) that allows for meaningful differentiation in school performance.¹⁰²

Based on the performance of all students and each subgroup of students, states must meaningfully differentiate school performance using all of the required indicators. States are required to identify (1) at least the lowest-performing 5% of all schools receiving Title I-A funds, (2) all public high schools failing to graduate one-third or more of their students, (3) schools required to implement additional targeted support (see below) that have not improved in a state-determined number of years, and (4) additional statewide categories of schools, at the state's discretion, for comprehensive support and improvement (CSI). States also are required to identify for targeted support and improvement (TSI) any school in which a subgroup of students is consistently underperforming. Schools in which one or more subgroups were performing at the same level as schools identified for CSI must be identified for additional targeted support and improvement (ATSI) activities. Thus, being able to disaggregate students by subgroup, including the economically disadvantaged student subgroup, is required for the identification of schools for improvement.

States and LEAs are required to prepare and disseminate annual report cards that include a range of information. LEAs are also required to prepare and disseminate report cards for each of their public schools. These report cards must include data disaggregated for economically disadvantaged students on student achievement on the mathematics, RLA, and science assessments required under Title I-A at each level of achievement¹⁰³; student performance on the other academic indicator included in the state's accountability system for elementary schools and secondary schools that are not high schools; high school graduation rates; student performance on the indicator(s) of school quality or student success used in the state's accountability system, as well as their progress toward meeting the state's long-term accountability system goals, including interim progress; and the percentage of economically disadvantaged students who were assessed and not assessed.

Thus, for outcome accountability and reporting purposes, schools need to be able to identify students in the economically disadvantaged subgroup. LEAs and schools have generally used FRPL data to comply with accountability and reporting requirements relating to subgroup performance for low-income students. As these data are no longer available in schools participating in CEP, ED has published guidance providing schools with a series of options for identifying students from low-income families (see below).

Data Options for LEAs Not Participating in CEP

The ESEA does not specify how to determine which students should be included in the economically disadvantaged subgroup. In practice, this subgroup has historically been based on FRPL data. For schools participating in the NSLP but not CEP, these data are still available.

Data Options for LEAs Participating in CEP

As FRPL data are no longer available for schools participating in CEP, ED has provided policy guidance that gives states and LEAs three options for determining low-income status for accountability and reporting purposes: (1) to consider all students in CEP schools to be from low-income families, (2) to consider only Identified Students to be from low-income families, and (3) to use income surveys to identify students from low-income families.¹⁰⁴ Each of these options, and its alignment with ESEA accountability and reporting requirements, is discussed in more detail below.

Assume All Students Are from Low-Income Families

LEAs may consider all students in CEP schools to be from low-income families. In this situation, the all students group is the same as the economically disadvantaged subgroup. ED has pointed out that the rate of students from low-income families is relatively high in CEP schools; thus, it is not unreasonable to consider all of the students in these schools as being from low-income families.¹⁰⁵ However, the rate of Identified Students in a CEP school can generally be as low as 40%.¹⁰⁶ Thus, it can be argued that although a school has a relatively high rate of Identified Students, the rate is not high enough to assume that 100% of students are from low-income families. Additionally, this approach obscures any achievement gaps for students from low-income families as it does not allow schools and LEAs to differentiate between all students and those from low-income families.

Consider Only Identified Students to Be from Low-Income Families

LEAs may consider Identified Students in CEP schools to be from low-income families. Unlike the previous option, this would most likely underrepresent the size of the economically disadvantaged subgroup, as Identified Students do not include all of the students who would be included in FRPL student counts.¹⁰⁷

One argument in favor of this approach is that, unlike the previous option, it allows schools and LEAs to differentiate between all students and those from low-income families. However, this approach may exclude students who previously would have been considered to be from low-income families because they would be eligible for FRPL based on a school lunch application.

Identify Students Based on Surveys

LEAs can use a household income survey to identify students from low-income families. As previously discussed, the data from a survey could be used for Title I-A purposes and for other programs. As with the previous option, income surveys allow schools and LEAs to differentiate between students from low-income families and the general student body. At the same time, income surveys essentially reintroduce some of the paperwork that CEP is intended to eliminate.

Alternative Ways to Measure School-Level Poverty or Related Indicators

Throughout the history of the Title I-A program, its focus has remained on providing funds to areas with concentrations of poverty. Thus, Congress has needed to identify which children should be considered as living in poor or low-income families. This has made it necessary to define poverty, identify a data source for measuring poverty, and decide which other categories of children, if any, should be included in the determination of Title I-A grants. These choices all have implications for state, LEA, and school grant amounts.

As discussed previously, data for the number of children in each school living in families in poverty are not readily available, so the number and percentage of children eligible to receive FRPL is often used as a proxy measure. According to the National Center for Education Statistics (NCES) at ED, "Because the free/reduced price lunch eligibility is derived from the federal poverty level, and therefore highly related to it, the free/reduced price lunch percentage is useful to researchers from an analytic perspective."¹⁰⁸ However, utility of the FRPL measure has changed substantially with the introduction of CEP, complicating its use as a measure to determine school-level grant amounts, meet Title I-A accountability and reporting requirements, and meet the needs of other programs or research that relies on FRPL data as a measure of school-level poverty.

ED summarizes issues related to the use of existing poverty measures and their limitations as such:

Although federal, state, and local education programs focus billions of dollars each year to improve educational opportunities for children in low-income schools and neighborhoods, the information available to identify and target high-need areas is limited. Most compensatory education programs use some type of poverty indicator to determine program eligibility and/or funding levels, but the development and accessibility of poverty data has not kept pace with the needs of these programs. The existing poverty thresholds do not fully reflect nonfood expenses needed to maintain household well-being, and they do not fully account for noncash in-kind benefits provided to individuals and families participating in federal need-based initiatives like the Supplemental Nutritional Assistance Program (SNAP) or the Women, Infants, and Children program (Citro and Michael 1995). Nor do they reflect important interactions between poverty status and other attributes like occupational prestige and educational attainment. More importantly, the structure and accessibility of poverty data is too limited. Carefully constructed measures of socioeconomic status (SES) provide little benefit if the resulting data are not available at the necessary geographic scale or for required analytic areas.¹⁰⁹

In response, ED has been studying possible alternative measures of school-level poverty as well as measures of SES that include components that address school-level poverty. These efforts are discussed below.

Alternative Measures of School-Level Poverty

ED currently is working on developing a new school-level poverty measure. In the request for applications for the Statewide Longitudinal Data Systems (SLDS) program released by the Institute of Education Sciences (IES) at ED in June 2019,¹¹⁰ ED indicated that applicants awarded grants under any of the three SLDS grant priorities (infrastructure, education choice, and equity) also were eligible to receive $250,000 to assist ED in testing a proposed school-level poverty measure. The new measure would be based on student addresses rather than FRPL eligibility. Participating states¹¹¹ would be required to create geocoded student address directories and to link these data with other geographic information provided by NCES. States would then use these data to produce summaries of the existing poverty measures based on FRPL and the proposed poverty measure to share with ED. Neither the geocoded student address directory created by a state nor individual student information would be shared outside the state. States would be required to participate in up to six webinars each grant year to discuss existing and proposed poverty measures. While a specific timeline for these activities was not specified, the SLDS grant awards are for 48 months, so it is possible that work on the development of a new school-level poverty measure will occur over this period of time as well.

NCES previously examined the feasibility of creating a flexible neighborhood poverty indicator that could be used to identify schools in low-income neighborhoods based on data from the ACS¹¹² and estimation techniques developed for spatial statistics.¹¹³ To determine what was involved with actually producing the proposed indicator, researchers developed neighborhood poverty estimates for almost 1,800 Ohio elementary schools.¹¹⁴ Based on the results of the study, the researchers determined the proposed indicator "may provide a useful supplement to existing school-level poverty indicators."¹¹⁵ They note there are many benefits to this approach, including the ability to create estimates for all schools, the use of a well-known and widely used poverty standard, and the use of data that "originate from a reliable, authoritative source that uses a consistent method of measuring income and poverty across the country."¹¹⁶ In addition, the data could be updated annually, would be "relatively cost-effective to produce," and would not suffer from disclosure limitations that often restrict the release of poverty estimates for small areas. The researchers also note the possible utility of using the estimates to determine Title I-A grants to the school level.

The proposed measures also have downsides. For example, the data only provide estimates for school neighborhoods, not for counties, cities, or congressional districts. They also do not provide the percentage of children in poverty actually enrolled in a school. In addition, the proposed indicator works less well for schools of choice that may draw students from many neighborhoods or for non-public schools that also enroll students from many areas. The researchers also note concerns about the sources of income included in the Census data used for the analysis and relatively large standard errors in the estimates for the new indicator. Despite these limitations, the researchers recommended the development of a set of indicators for all public schools for further use and study.

ED also has examined the utility of creating school-level poverty estimates using data from the ACS, which are used to develop the SAIPE data employed in determining Title I-A LEA grant amounts. As part of the NCES School Attendance Boundary Survey (SABS),¹¹⁷ NCES examined whether it was possible to "integrate school attendance boundaries with data from the ACS to develop demographic estimates for individual school areas."¹¹⁸ While NCES determined that estimates could be created, the "average quality for these small geographic areas was too unreliable for NCES to create and release as a regular public data product." NCES also identified challenges related to how often school area boundaries were updated and to schools with open-enrollment policies that may not have neighborhood boundaries.

Alternative Measures of Family Characteristics That May Be Related to Student Achievement

While FRPL data can serve as an indicator of relative poverty, they are not actual measures of poverty or changes in poverty rates. In addition, neither FRPL nor poverty measures are measures of socioeconomic status (SES, which "measures a broader spectrum of family characteristics [e.g., parental education and occupations] that may be related to student performance"¹¹⁹). This section of the report discusses ongoing efforts at ED to develop SES measures that would include components that could be used to measure family income.

In 2012, in response to a request for the National Assessment Governing Board (NAGB), NCES convened an expert panel to examine ways in which the measurement of SES for purposes of the National Assessment of Educational Progress (NAEP) could be improved. The panel reached consensus on the following definition of SES:

SES can be defined broadly as one's access to financial, social, cultural, and human capital resources. Traditionally, a student's SES has included, as components, parental educational attainment, parental occupational status, and household or family income, with appropriate adjustment for household or family composition. An expanded SES measure could include measures of additional household, neighborhood, and school resources.¹²⁰

NCES also established an Alternative Socioeconomic Status Measures Working Group (hereinafter, Working Group), which was part of the National Forum on Education Statistics.¹²¹ In response to the potential loss of the NSLP-eligibility indicator, the group was tasked with "identifying alternative measures of SES that meet the needs of the education community."¹²² The working group was composed of staff from NCES and ED as well as state and LEA staff.

In 2015, the Working Group published the Forum Guide to Alternative Measures of Socioeconomic Status in Education Data Systems.¹²³ This document was created to provide information to the education community as it considers alternatives to FRPL data as a proxy for student and family SES. The guide highlights three challenges with continuing to use FRPL data. First, FRPL data are being used and interpreted in ways that were not intended by the data's collection. FRPL is used as a proxy for SES even though it only measures family income and does not include other relevant measures such as parents' occupation and education. Second, FRPL data at the individual level are available only for purposes of administering the school meals program; "NSLP prohibits other education staff from using the data to determine the instructional and non-instructional (service eligibility) needs of individual students."¹²⁴ Third, with the increasing use of CEP, the availability of individual student-level FRPL data is declining in participating LEAs.

The Working Group focuses on eight alternative SES measures

• 1. eligibility for other means-tested programs,
• 2. household-provided information,
• 3. student/family categorical status (e.g., homeless, migrant, foster care, runaway),
• 4. family/household income,
• 5. highest level of education completed by resident parent/guardian,
• 6. occupation of resident parent/guardian,
• 7. neighborhood SES, and
• 8. school district poverty estimate.

While the Working Group notes that any of these components could be used as a stand-alone proxy for SES, it does not recommend this approach. Rather, it recommends using the measures in combination with one another. According to the Working Group, when measures 1-3 are combined, they replicate past methods of identifying FRPL eligible students and, therefore, may be consistent with historical values. Measures 4-6 "reflect the three components of socioeconomic status commonly recognized by the research community."¹²⁵ The last two measures describe community-related aspects of an individual's SES.

Current and Next Steps on Measures of School-Level Poverty

As discussed previously, Title I-A always has relied on measures of family income to determine grant amounts. In addition, at the school level Title I-A grants have historically been made and continue to be made to schools with relatively high concentrations of students from low-income families. According to ED:

The impact of poverty on student achievement, educational attainment, and other educational outcomes has been a concern for educators and federal policymakers since the passage of the Elementary and Secondary Education Act (ESEA) in 1965. Educational programs like Title I, Head Start, Promise Neighborhoods, E-Rate, and the National School Lunch Program (NSLP) target federal resources to help mitigate the effects of poverty on low-income students, families, and neighborhoods.¹²⁶

Title I-A is not the only ESEA program that uses measures of poverty to make grants at the state or LEA level. For example, there are two formula grant programs authorized by the ESEA under which grants are made to states and subsequently to LEAs in proportion to prior-year Title I-A grants—Student Support and Academic Enrichment Grants program (Title IV-A) and 21^st Century Community Learning Centers (Title IV-B).¹²⁷ Other programs also use measures of poverty to determine grant amounts or eligibility to receive a grant. The Supporting Effective Instruction program (Title II-A) uses a combination of Census population data for children ages 5-17 and SAIPE poverty estimates for children ages 5-17 to determine grants to states and subgrants to LEAs. Similarly, under the Rural and Low-Income Schools (RLIS) program (Title V-B-2), LEAs are eligible for a grant if, among other eligibility requirements, at least 20% of the children ages 5-17 served by the LEA are from families with income below the poverty line, which can be determined using SAIPE data.

Title I-A is unique among federal education programs administered by ED with respect to its statutory requirements related to the distribution of funds to the school level. It also includes, as a condition of accepting Title I-A funds, numerous accountability and reporting requirements for states, LEAs, and schools that require the disaggregation of data by low-income student status. As FRPL data become less useful for meeting these needs and the needs of other programs or research that may also rely on FRPL data in some capacity (e.g., reduction in school fees for students who are FRPL eligible, philanthropic organizations that use FRPL data in grant determinations), this raises the issue of what school-level poverty measure could be used instead of FRPL data.

While there is continued congressional interest in using poverty measures to determine grant allocations under federal education programs, those that are currently being used are imperfect as measures of poverty and, some may argue, are too limited in their focus. As discussed above, ED currently is engaged in a study with multiple states to develop a new school-level measure of poverty. Should this measure come to fruition, it could be used to make Title I-A grants to schools, for Title I-A accountability purposes, and for research requiring school-level measures of poverty. In the meantime, ED notes that despite the limitations and data quality issues associated with the FRPL counts, they "continue to serve as the standard for identifying school-level poverty for educational programs and surveys because they satisfy core conditions needed to serve as useful program indicators."¹²⁸ Drawing on the work of the researchers Harwell and LeBeau,¹²⁹ ED lists these conditions and describes how FRPL data meet the criteria of universal participation, uniform criteria, regular updates, stable infrastructure, flexible application, easy access, and cost-effective development.¹³⁰ With respect to the last condition, ED points out that the development of a new school-level poverty measure that requires the collection of new data from all schools in the United States would be cost prohibitive and would require new statutory authority and new funding. ED notes that trying to repurpose existing data would be a more "convenient and cost-effective solution."

Brief Considerations for Congress

As previously discussed, LEAs currently lack a consistent measure to identify low-income children across schools for determining Title I-A school-level allocations and for academic accountability purposes under the ESEA. Congress could consider requiring ED to report on its previous and current efforts to identify a new measure of low-income students that would be both reliable and consistent across schools. In addition, Congress could consider engaging with ED to determine whether a better measure could be developed if new statutory authority and new funding were provided and what specific authority and funding levels would be needed to create a new measure.

Appendix A. Overview of Title I-A Formula Factors

Appendix B. Current Issues Regarding the Standard Federal Poverty Measure as Applied in the SAIPE Estimates Used to Calculate Title I-A Grants to LEAs

Since adoption of the Education Amendments of 1974 (P.L. 93-380), ESEA Title I-A grants have been determined by estimates of the number of related children ages 5-17 in families with income below the standard federal poverty measure's thresholds. This Appendix discusses the development and characteristics of this measure of poverty, as well as major proposals to modify the measure over recent decades—including a Supplemental Poverty Measure (also published by the Census Bureau).

Characteristics of the Standard Federal Poverty Measure¹³¹

As noted above, the standard federal poverty measure has been the criterion for determining poverty level family income, as applied in the allocation of Title I-A grants to states and LEAs, for statistical purposes. Thus, all of the poverty estimates provided by the SAIPE program, and used to calculate state and/or LEA grants under Title I-A and other ESEA programs, are based on this standard for determining whether school-age children are in families that have income that is below the poverty level. In addition, standards for determining whether students are eligible for FRPL, or whether their families are eligible for such programs as SNAP, TANF, or Medicaid are also based on the same standard federal poverty measure. Therefore, this measure is a critical underlying factor in all aspects of fund allocation under Title I-A and several other ESEA programs.

The measure, initially developed in the early 1960s, and adopted as the standard federal poverty measure in 1969, is sometimes called the Orshansky measure, after its developer, Mollie Orshansky, an economist working for the Social Security Administration. The system for setting the standard federal poverty income thresholds was originally based on survey data indicating that low-income families spent one-third of their income on food. Thus, poverty income thresholds were set, in general, at three times the estimated cost of food to meet minimally adequate dietary guidelines, based on a survey conducted in 1955.¹³² Then, as now, the standard is based solely on gross cash income. Different income thresholds were established based on family size, number of children in the family, the sex of the head of the household, and farm versus non-farm residence. Beginning in 1980, the number of different income thresholds was reduced to consider only two categories—family size and number of children.¹³³ For two-person households only, the thresholds also vary depending on whether the household includes an adult whose is 65 or older.

The standard federal poverty income thresholds are updated annually, based on annual changes in the Consumer Price Index for Urban Consumers (CPI-U).¹³⁴ The most recent published poverty income thresholds, applicable to income for calendar year 2019, are listed in Table B-1, below.

Concerns About the Standard Federal Measure of Poverty: "Measuring Poverty: A New Approach"

A major study of the current federal poverty standard, with recommendations for changes to it, was published by the National Research Council (NRC) of the National Academy of Sciences in 1995. This report, titled "Measuring Poverty: A New Approach,"¹³⁵ critiqued the then (and still) current standard federal measure of poverty, and recommended a number of changes to it.¹³⁶ The report focused on defining a threshold or budget level below which families would be considered poor, and the level of income or resources available to families to compare with that threshold.

The report addressed numerous issues and concerns that had arisen regarding the standard federal poverty measure in the decades following its development. These include the following:

• The standard measure was based on a 1955 survey of family expenditure patterns. Since that time, family expenditure patterns have changed significantly. For example, the share of family budgets devoted to food costs has declined, while the share devoted to certain other basic necessities, such as housing or medical costs, has increased.
• The standard measure is based on gross cash income, and therefore does not take into consideration the cost of income, payroll (Social Security), and other taxes that families must pay from that income. The standard measure also does not consider the value of tax benefits that families may receive, such as under the Earned Income Tax Credit (EITC) or the Child Care Tax Credit, especially in cases where those tax credits are refundable.¹³⁷
• Because it is based on gross cash income, the standard measure considers only benefit programs that provide pre-tax cash income. These include programs such as Social Security, TANF (previously, AFDC), and Supplemental Security Income (SSI). Not considered are programs that provide non-cash benefits, such as SNAP (previously, Food Stamps), Medicaid, or housing subsidies.
• There is no geographic adjustment to the standard thresholds. That is, the same thresholds are applied to households throughout the nation, regardless of differences in the cost of living in different areas.
• The standard thresholds are adjusted annually with respect to a broad measure of inflation in consumer prices (CPI-U), but this may not accurately reflect changes in prices for basic necessities for low-income families.
• The standard measure does not consider differences in health status, health insurance coverage, or out-of-pocket medical costs.

In response, the NRC panel recommended a number of changes to the income thresholds associated with the standard federal measure of poverty. The most general aspects of the proposed changes regarding poverty income thresholds included the following:

• The thresholds should be based on current estimated needs and consumption patterns for food, housing (including utilities), and clothing, plus an amount for miscellaneous expenses (household supplies, personal needs, transportation, etc.).
• The thresholds should be adjusted annually on the basis of specific measures of changes in the costs of food, housing, and clothing, not the broad CPI-U measure, and adjusted over time to consider future changes in patterns of consumption of these items.
• The thresholds should be adjusted for differences in costs in different areas of the nation, especially (but not only) with respect to housing.

With respect to the family income or resource levels that should be compared to the above thresholds, in order to determine the poverty status of the family or household, the NRC recommended these should be based on money and near-money disposable income. Thus, they recommended that family income should be reduced with respect to

• income and payroll taxes;
• employment expenses, such as child care where necessary;
• out-of-pocket medical costs, including those for health insurance premiums; and
• child support payments.

The NRC panel recommended that family income levels should be increased with respect to

• a wide variety of in-kind benefits, including those provided under SNAP (previously, Food Stamps), housing and energy subsidy programs, and child nutrition programs; and
• tax benefits, particularly the EITC and child care credits.

Alternative Measure of Poverty Published by the Census Bureau: The Supplemental Poverty Measure¹³⁸

Since 2011, the Census Bureau has published a series of reports containing data on the number and characteristics of people in poverty determined by applying a Supplemental Poverty Measure (SPM).¹³⁹ The SPM was developed by an Interagency Working Group on Developing a Supplemental Poverty Measure (ITWG) that included staff from a range of federal agencies. The SPM incorporates many of the recommendations of the NRC in the 1995 report discussed above. According to the Census Bureau, the SPM is not intended to be a replacement for the standard federal poverty measure, or to be implemented in the administration of federal tax or benefit program policies; it is intended to help illustrate the effects on the population in poverty of current federal policies in these areas.

Major differences between the standard federal poverty measure and the SPM are summarized in Table B-2.

As shown in Table B-2, the SPM applies different income thresholds, somewhat different units of analysis, and different mixes of resources to the estimation of the population in poverty. According to the most recent report on the SPM, based on income for calendar year 2018, the overall population poverty rate was slightly higher using the SPM (12.8%) than the standard federal poverty measure (11.8%). However, application of the SPM versus the standard measure varies with respect to a number of population characteristics, as illustrated in Table B-3.¹⁴⁰

As indicated in Table B-3, while the overall poverty rate for 2018 was 1.0 percentage point higher under the SPM than under the standard poverty measure, there were substantial variations among people in different age groups, regions, and metro-/non-metro areas of the nation.¹⁴¹ Among age groups, the SPM rate of 13.7% for children from birth through age 17 was substantially lower than the standard measure rate of 16.2%, in part because of a number of benefit programs focused on families with children (school nutrition programs, WIC, etc.) that are considered non-cash income when determining if the SPM threshold is met. Regarding the effects of specific adjustments to income under the SPM on the poverty rates for children (ages birth through 17), the adjustments that had the greatest effect on reducing poverty rates were those for refundable tax credits (EITC and the refundable portion of the child care tax credit), SNAP, and Social Security payments. The adjustments that had the greatest effect of increasing poverty rates for 0-17 year olds were out-of-pocket medical expenses, work-related expenses, and Social Security payroll taxes.¹⁴² The SPM rate of 13.6% for adults aged 65 and older was higher than the standard measure rate of 9.7%, in part because of the higher out-of-pocket costs for medical care among seniors.

Among regions of the nation, the SPM rate was higher than the standard measure rate for states in the Northeast (12.2% vs. 10.3%) and West (14.4% vs. 11.2%), in part due to higher costs of housing in those regions. The SPM poverty rate was lower than the standard measure rate for states in the Midwest (9.2% vs. 10.4%), and the two poverty rates were similar for states in the South (SPM 13.9%, standard measure 13.6%).

For people living in metropolitan versus non-metropolitan areas of the nation, the SPM rate was higher than the standard measure rate for all persons living in Metropolitan Statistical Areas (MSAs) (12.9% vs. 11.3%) and specifically those living in principal cities of MSAs (16.0% vs. 14.6%), while the SPM rate was lower for those living outside MSAs (12.2% vs. 14.7%). This is in part due to the SPM adjustments for differences in housing costs.