Reasons for Measuring Poverty in the United States in the Context of Public Policy — A Historical Review, 1916-1995. The Seventies and the Eighties


In 1974, in response to a legislative requirement, an interagency Poverty Studies Task Force was established under the leadership of the U.S. Department of Health, Education, and Welfare to undertake an intensive review of the current poverty measure and various alternative measurement schemes. The Task Force's final report83 was submitted to Congress in April 1976. This report thoroughly explored the issues involved in developing and revising poverty measures, gathering extensive supporting information which was presented in the report itself and in 17 Technical Papers. The report did not recommend specific changes in the current poverty measure.84 The report's Forwarding Letter alluded to the use of the poverty measure in the fund allocation formula for Title I of the Elementary and Secondary Education Act of 1965, and also alluded to "administrative, analytical, statistical, and budgetary implications of alternative [poverty] measurement schemes."85 The Executive Summary of the report noted that "[t]his country is concerned about poverty, its causes and correlates. It is willing to relieve the poverty of some of the poor and it wants to measure the effectiveness of its efforts to do so. None of this can be done without some idea of who is to be considered poor and who is not.... The Orshansky poverty definition is widely used to measure the nation's progress in reducing the extent of poverty. It is also used as a statistical tool to identify the target populations of government programs that help the financially needy and to evaluate the effectiveness of such programs. In recent years it has been adapted for administrative purposes." (In this connection it mentioned use of the poverty measure both in fund allocation formulas [the poverty thresholds] and in the determination of eligibility for programs [the poverty guidelines].) The Executive Summary concluded with a statement that "there is an advantage in the continued publication of an official statistical series of a poverty measure as an index of national achievement in reducing the extent of poverty."86 The Introduction included a statement that "[p]overty can be defined in the abstract but whether someone is to be considered poor ultimately depends on who is asking the question and why he [sic] wants to know."87 The first chapter of the report, in discussing arguments for using an official poverty measure, commented that "[d]etermining public policy, setting program goals, and evaluating program success all benefit from the external referent provided by an official [poverty] measure, even if the measure is altered to fit programs. It allows comparisons over time. Also, various Federally funded programs are required by law to identify their target populations in terms of low-income status or extent of poverty...." It noted that an official poverty measure could be used "as an index of national achievement, as a source of statistical data for the design and evaluation of programs, and as an administrative parameter in Federal programs."88

In January 1977, the U.S. Congressional Budget Office (CBO) issued a background paper89 estimating the number of families in poverty (using the existing poverty thresholds without any alteration) under alternative definitions of income — in particular, under a definition of income that counted major government noncash benefits as income at full government cost. (In June 1977, CBO issued a corrected version90 of the background paper to correct a computer programming error.) The analysis had been requested in May 1976 by then-Senator Walter Mondale [D-Minn.]; Senators Edmund Muskie [D-Me.] and Henry Bellmon [R-Okla.] had also expressed interest in the preparation of the study. The paper noted that "[t]he 95th Congress will be considering legislation to reform social welfare programs. Important criteria for evaluating any new proposal are how it will affect families in poverty and what it costs. This paper provides the basis for such an evaluation by analyzing how the current income transfer programs lift families out of poverty.... During the past decade, public expenditures for social welfare programs have grown four-fold — from $77.2 billion in 1965 to $286.5 billion in 1975. At the same time, according to official poverty statistics, the percentage of families in poverty has declined by only 30 percent." [The social welfare expenditures given are in current dollars, whereas poverty is measured by constant-dollar thresholds. Furthermore, not until the next paragraph was it noted that a majority of social welfare transfer programs are directed to the general population, not just the poor.] "An apparent paradox, this situation has led some observers to question the efficacy of the current system of public transfers. This dilemma is the result of two factors: the types of programs that account for most of the recent growth; and the inadequacies of the measures used to estimate families in poverty."91 The analysis adjusted Current Population Survey income data for underreporting and nonreporting, took out [estimated] taxes, and used a microsimulation model to impute recipiency of major government noncash benefits (as well as Aid to Families with Dependent Children and Supplemental Security Income) to families. On that basis, the analysis estimated that the number of family units in poverty in fiscal [sic] year 1976 was 10.7 million on the basis of money income only, but only 6.6 million after counting noncash benefits as income and taking out taxes.92

In March 1982, the Census Bureau issued Technical Paper 5093 — its first report on the valuation of noncash benefits and poverty. This "exploratory" report included a statement that "[t]his research report is guided by requests from Congress and the Office of Management and Budget...for such an analysis."94 The report's first appendix was a "U.S. Senate statement" (more precisely, language inserted by the Senate Appropriations Committee into a 1980 report on the Committee's version of an appropriations bill) "expressing their concern over this issue."95 The Senate Appropriations Committee had inserted this language at the request of Senator Henry Bellmon (R-Okla.)96; a few months earlier, Senator Bellmon had put a statement97 on valuing in-kind benefits into the Congressional Record. Accordingly, not only Technical Paper 50 but also the Senate Appropriations Committee language and Senator Bellmon's Congressional Record statement might be able to shed some light on the reasons for the "poverty" measures (actually different income measures applied to an unchanged poverty measure) in Technical Paper 50.

In his June 1980 Congressional Record statement, Senator Bellmon included the summary of a paper by William Hoagland which counted in-kind transfers as income for poverty measurement purposes, reducing the projected 1980 poverty count from 19 million persons (under the current income definition) to 13 million (not counting medical benefits as income) or 9 million (counting medical benefits as income).98 The Senator commented that "...poverty is still a serious problem in this Nation, but it is not as big a problem as it once was.... Billions of dollars of benefits under current Government transfer programs are simply ignored in preparing official estimates of the numbers of people living in poverty. The growth of in-kind benefit programs makes it essential that we begin to take these benefits into account in measuring personal incomes. Otherwise, our decisions on tax and spending proposals affecting the poor and near-poor will be based on incomplete and potentially misleading information."

The September 1980 Senate Appropriations Committee language said that the Census Bureau's official poverty statistics "ignore billions of dollars of Government in-kind benefits," and mentioned the 9 million [projected] poverty count from the Hoagland paper. "The official statistics show no significant reduction in recent years in the incidence of poverty, although in-kind benefit programs have expanded greatly. The Committee considers it essential that official poverty statistics reflect, at the earliest possible date, the effects of in-kind benefits. Without such information Congress and the Executive Branch cannot be certain that Government transfer programs are properly targeted."

According to Technical Paper 50 itself, "[t]he purpose of this report is to examine several alternative methodologies for valuing public in-kind (noncash) transfers and to assess their effect on the size and composition of the official poverty population. The current definition of poverty used for statistical based on money income and does...not include the value of in-kind transfers as income. Between 1965 and 1980, the market value of major in-kind transfers in the form of food...housing...and medical care (Medicare and Medicaid) grew from $2.2 billion to over $72.5 billion. In-kind transfers intended for the low-income population currently exceed cash public assistance by more than two to one. The Government's statistics on poverty have been criticized by a number of experts because they fail to account for noncash benefits.... The period between 1965 and 1979 saw sharply declining poverty rates during the first 5 years followed by a 10-year stretch in which poverty declined only slightly....using the present definition of income. Given these trends, widespread public interest concerning the disposition of these [in-kind public] transfer payments and their effect on the poor has arisen during the last decade.... Unfortunately, most research studies [on valuation of in-kind transfers] have not established the conceptual basis for their approach to valuing in-kind transfers, and the use of inappropriate measures of the value of noncash benefits may lead to incorrect policy decisions."99 The report presented three different approaches to valuing selected government noncash benefits. For each approach, the report presented three estimates — one valuing food and housing benefits only, one valuing food and housing benefits and medical benefits excluding institutional care, and one valuing food and housing benefits and medical benefits including institutional care; this was done because of "the importance of medical benefits, and...the special problems in valuing medical care benefits, especially institutional care benefits," as well as conceptual questions about the appropriateness of including medical benefits in the determination of poverty status.100 The report made use of Current Population Survey data on actual receipt of or coverage by various noncash benefits, rather than using microsimulation models to impute recipiency of benefits to households.101 Among the various valuation approaches and combinations of noncash benefits valued, the smallest reduction in counted poverty resulted from valuing food and housing benefits only under the "poverty budget share" approach (count reduced from 24 million persons in 1979 based on money income only to 21 million persons), and the largest reduction resulted from valuing food and housing benefits and medical benefits including institutional care under the "market value" approach (count reduced to 14 million persons).102

83.  U.S. Department of Health, Education, and Welfare, The Measure of Poverty[:] A Report to Congress as Mandated by The Education Amendments of 1974, Washington, D.C., [U.S. Government Printing Office,] April 1976.

84.  For further discussion of the Poverty Studies Task Force and its report, see Fisher, "The Development of the Orshansky Poverty Thresholds..." (1997 revision), pp. 38-40.

85.  The Measure of Poverty..., p. iii.

86.  The Measure of Poverty..., pp. xxi and xxvi; see also p. 1.

87.  The Measure of Poverty..., p. 2.

88.  The Measure of Poverty..., p. 17.

89.  Congressional Budget Office, Poverty Status of Families Under Alternative Definitions of Income (Background Paper No. 17), Washington, D.C., U.S. Government Printing Office, January 13, 1977. See also Peter Milius, "Study Finds 50% Fewer Poor People[:] Non-Cash Benefits Counted as Income In Report for Hill," Washington Post, January 16, 1977, pp. A1 and A17.

90.  Congressional Budget Office, Poverty Status of Families Under Alternative Definitions of Income (Background Paper No. 17 (Revised)), Washington, D.C., U.S. Government Printing Office, June 1977 (Reprinted October 1977). "The revision was necessary to correct a computer programming error that resulted in a systematic underestimate of the number of families in poverty under all definitions of income. Although all of the poverty incidence figures were affected, the correction of this error does not affect the findings of the earlier report" (p. iii).

91.  Background Paper No. 17 (Revised), pp. iii and xiii; see also pp. 1-3. Note that on p. xiii of the background paper, John Korbel (author of the background paper, per p. iii) seems to have used an updated version of a piece of "floating rhetoric" used several years earlier by conservative Edgar Browning: "Total social welfare expenditures increased from $77.2 billion in 1965 to $215 billion in 1973, or by about 180 percent. Despite this vast effort, the number of people officially defined as poor fell by less than a third, from 33.2 million to 23 million" (Edgar K. Browning, Redistribution and the Welfare System (Evaluative Studies 22), Washington, D.C., American Enterprise Institute for Public Policy Research, July 1975, p. 7).

92.  Background Paper No. 17 (Revised), pp. xiv-xv, 7-9, and 17-19.

93.  U.S. Bureau of the Census, Technical Paper 50, Alternative Methods for Valuing Selected In-Kind Transfer Benefits and Measuring Their Effect on Poverty (by Timothy M. Smeeding), Washington, D.C., U.S. Government Printing Office, March 1982.

94.  Technical Paper 50, p. 1, footnote 2. For the "very exploratory" or "highly exploratory" nature of the report, see pp. v and 1.

95.  For the quoted description of the Senate Appropriations Committee language, see Technical Paper 50, p. 1, footnote 2. For the appendix in question, see "Appendix A. U.S. Senate Statement: 'Data Collection and [sic] Poverty Level,'" p. 105 of Technical Paper 50. For the original source of the Senate Committee language, see U.S. Senate, Committee on Appropriations, "Data Collection on Poverty Level," pp. 33-34,Departments of State, Justice, and Commerce, the Judiciary, and Related Agencies Appropriation Bill, 1981 [Report to accompany H.R. 7584], Senate Report No. 96-949, 96th Congress, 2d Session, Washington, D.C., September 16, 1980. For similar language included in the conference report by the House-Senate conference committee on the appropriations bill, see U.S. House of Representatives, "BUREAU OF THE CENSUS[:] Salaries and Expenses," pp. 8-9, Making Appropriations for the Departments of State, Justice, and Commerce, the Judiciary, and Related Agencies [Conference Report to accompany H.R. 7584], House Report No. 96-1472, 96th Congress, 2d Session, Washington, D.C., November 20, 1980. Technical Paper 50 did not include the text of the Office of Management and Budget request (a letter from an OMB official to the Chief Economist of the Commerce Department — the "Cutter..." bibliographic entry on p. 157 of Technical Paper 50).

96.  Gary D. Bass (Executive Director, OMB Watch), pp. 5 and 14 in U.S. House of Representatives, Census Bureau Measurement of Noncash Benefits[:] Hearings Before the Subcommittee on Census and Population of the Committee on Post Office and Civil Service..., Serial No. 99-51, Washington, D.C., U.S. Government Printing Office, 1986.

97.  Senator Henry Bellmon [R-Okla.], "The Current Effectiveness of Current Federal Government Transfer Programs in Reducing Poverty" [Additional Statement — June 12, 1980], Congressional Record, Vol. 126, Part 11 (96th Congress, 2d Session), pp. 14450-14451, Washington, D.C., U.S. Government Printing Office. Note that in 1971, Senator Bellmon had put a statement into the Congressional Recordcriticizing the poverty measure for not counting assets or income received from sale of property, loans, gifts, and lump-sum inheritances and insurance payments. (See Senator Henry Bellmon, "Need for a Meaningful Poverty Definition" [Additional Statement — May 19, 1971],Congressional Record, Vol. 117, Part 12 (92d Congress, First Session), p. 15860, Washington, United States Government Printing Office, 1971.)

98.  For the published version of Hoagland's paper, see G. William Hoagland, "The Effectiveness of Current Transfer Programs in Reducing Poverty," pp. 53-75 in Paul M. Sommers (editor), Welfare Reform in America[:] Perspectives and Prospects, Boston, Kluwer. Nijhoff Publishing, 1982. Hoagland projected poverty figures for Fiscal Year rather than calendar year 1980. He adjusted Current Population Survey data for income underreporting and nonreporting — presumably the main reason for his surprisingly low poverty figure under the current income definition. The actual calendar year 1980 poverty count (under the current income definition, without adjustment for income underreporting and nonreporting, and before minor revisions in the poverty definition) was 29.3 million persons.

99.  Technical Paper 50, pp. v, 4, and 26.

100.  Technical Paper 50, pp. vi and 5 (sources of the quotations) and 72-75.

101.  Technical Paper 50, pp. 4, 92, and 94. For a discussion of how microsimulation models apparently overestimate noncash benefit recipiency in general and multiple benefit recipiency in particular, see Timothy M. Smeeding, "Estimating In-Kind Income Using the CPS: Survey vs. Pure Microsimulation Approaches," American Statistical Association[:] 1981 Proceedings of the Section on Survey Research Methods, pp. 360-365.

102.  Technical Paper 50, pp. vii, ix, and 80.