There has been renewed interest in the United States in recent years in issues relating to the definition and measurement of poverty.⁽¹⁾ In October 1989, the Joint Economic Committee of Congress released a staff study discussing current poverty measurement procedures and suggesting that the poverty thresholds be raised in real terms to reflect the major changes in consumption patterns and relative prices that have occurred in the United States since the mid-1950's. In January 1990, the Bush Administration approved an initiative on improving the quality of federal economic statistics; the current poverty measure was one of several dozen statistical series which was examined as part of that initiative. In April 1990, Urban Institute economist Patricia Ruggles published a book⁽²⁾ which urged an increase in the poverty line to reflect changes in consumption patterns and changing concepts of what constitutes a minimally adequate standard of living. In July 1990, two private organizations concerned with the poor and the elderly issued a report⁽³⁾ reviewing current poverty measurement procedures and describing a Gallup poll in which a nationally representative sample of Americans set an average dollar figure for the poverty line which was 24 percent higher than the current official poverty line. In 1992 and 1993, two pairs of authors published budgets⁽⁴⁾ developed with the purpose of replacing the current poverty thresholds. And in May 1995, a Panel on Poverty and Family Assistance appointed by the National Academy of Sciences/National Research Council (in response to a 1990 Congressional request) published a report⁽⁵⁾ proposing a new approach for developing an official poverty measure for the U.S.

In view of this recent renewed interest in poverty definition and measurement, it may be useful to reexamine how the current official poverty thresholds were originally developed, and what their subsequent history has been. The following account of the poverty thresholds is based largely on primary sources--e.g., papers and articles by Mollie Orshansky, who developed the thresholds, and unpublished records of federal interagency committees which reviewed the thresholds at various times. (A number of published descriptions of the poverty thresholds have either failed to consult published primary sources with sufficient care or have failed to consult them altogether, resulting in errors and inaccuracies.) This account concentrates on the internal or administrative history of the poverty thresholds; external debates about the poverty measure have been conducted in large part on the public record, while the primary sources for the internal administrative history tend to be either neglected or simply not available in published form.

The poverty thresholds are the primary version of the federal poverty measure--the other version being the poverty guidelines. The poverty thresholds are issued nowadays by the Census Bureau, and are generally used for statistical purposes-- for example, for estimating the number of persons in poverty nationwide each year and presenting data classifying them by type of residence, race, and other social, economic, and demographic characteristics. The poverty guidelines⁽⁶⁾ are issued by the Department of Health and Human Services, and are used for administrative purposes--for instance, for determining whether a person or family is financially eligible for assistance or services under certain federal programs.

The poverty thresholds were originally developed in 1963 and 1964 by Mollie Orshansky, an economist working for the Social Security Administration (SSA). As indicated below, she actually developed two sets of poverty thresholds--one derived from the Agriculture Department's economy food plan and one derived from the Agriculture Department's somewhat less stringent low-cost food plan. She described an initial version of these thresholds--for families with children only--in a July 1963 article in the Social Security Bulletin.⁽⁷⁾ She published an analysis using a refined and extended version of the two sets of thresholds (including thresholds for unrelated individuals and families without children, as well as for families with children) in a January 1965 Social Security Bulletin article.⁽⁸⁾

As Orshansky later indicated, her original purpose was not to introduce a new general measure of poverty⁽⁹⁾; instead, she was trying to develop a measure to assess the relative risks of low economic status (or, more broadly, the differentials in opportunity) among different demographic groups of families with children.⁽¹⁰⁾ However, her work appeared at a strategic time. The Johnson Administration had announced a War on Poverty in January 1964, and in late 1964 (when preprints of Orshansky's January 1965 article were being widely circulated) the Economic Opportunity Act of 1964 was being implemented and the Office of Economic Opportunity (OEO) was being set up.⁽¹¹⁾ As noted below, OEO adopted the lower of Orshansky's two sets of poverty thresholds as a working definition of poverty for statistical, planning, and budget purposes in May 1965. Certain aspects of the poverty thresholds were revised in August 1969 based on the recommendations of a federal interagency committee; in the same month, the Bureau of the Budget designated the just-revised poverty thresholds as the federal government's official statistical definition of poverty.⁽¹²⁾ Additional minor revisions in the poverty thresholds were made in 1981 based on the recommendations of another federal interagency committee.

This paper will first describe how Orshansky developed the poverty thresholds. It will then describe some of the background and events which led up to Orshansky's 1963 development and 1964 refinement and extension of the thresholds. (The chronological sequence is partially reversed in this fashion because some of the earlier events are easier to understand when placed in the context of how the thresholds were developed.) It will then describe the history of the thresholds after the Office of Economic Opportunity and other federal agencies started using them in 1965; this last part will focus to a great extent on reviews of the thresholds by various federal interagency committees.

How Orshansky Developed the Poverty Thresholds

Orshansky did not develop the poverty thresholds as a standard budget--a more precise technical term for what is today commonly called a "market basket." (A standard budget is "a list of goods and services that a family of a particular size and composition would require in a year to live at some specified level."⁽¹³⁾) If generally accepted standards of minimum need had been available for all or most of the major essential consumption items of living--housing, medical care, clothing, transportation, and so on--Orshansky could have followed a standard budget approach by costing out all the standards and adding up the costs. However, except for the area of food, no definitive and accepted standards of minimum need for major consumption items existed at the time Orshansky developed the thresholds--and it is still true that no such standards in non-food areas exist today.⁽¹⁴⁾ It should be noted that Orshansky's decision not to use the standard budget approach in developing the poverty thresholds was based on considerably more familiarity with standard budgets⁽¹⁵⁾ than many "mainstream" economists had at the time (or have today). While her 1959 article on standard budgets⁽¹⁶⁾ was written well before she began to develop the poverty thresholds, its discussion of some of the issues and problems with such budgets (pp. 12-13, 17) can be read--strictly in hindsight-- almost as a listing of reasons for not using the standard budget approach in developing the thresholds.⁽¹⁷⁾

In Orshansky's words, "...there is no generally accepted standard of adequacy for essentials of living except food."⁽¹⁸⁾ As her "generally accepted" standards of adequacy for food, she made use of the food plans prepared by the Department of Agriculture.⁽¹⁹⁾ At the time she was developing the thresholds, the Agriculture Department had food plans at the following four cost levels (listed here from the most costly to the cheapest): liberal, moderate, low-cost, and economy. The first three plans go back to the original introduction of the food plans in 1933⁽²⁰⁾, while the economy food plan was developed and introduced in 1961 using data from the Agriculture Department's 1955 Household Food Consumption Survey.⁽²¹⁾ Orshansky used the low-cost and economy⁽²²⁾ food plans in developing her two sets of poverty thresholds, describing them as follows: "The low-cost plan, adapted to the food patterns of families in the lowest third of the income range, has for many years been used by welfare agencies as a basis for food allotments for needy families and others who wished to keep food costs down. Often, however, the actual food allowance for families receiving public assistance was less than that in the low-cost plan....spending as much as this food plan recommends by no means guarantees that diets will be adequate....Recently the Department of Agriculture began to issue an 'economy' food plan, costing only 75-80 percent as much as the basic low-cost plan, for 'temporary or emergency use when funds are low.'...The food plan as such includes no additional allowance for meals eaten out or other food eaten away from home."⁽²³⁾ To be more precise, what Orshansky used in developing the poverty thresholds was the dollar costs of the foods in the two food plans; while the actual foods in both plans provided a fully nutritious diet, families spending for food at the dollar cost level of the economy food plan "had about an even chance of providing a fair or better diet for [the] family, but really only one chance in 10 of providing a good diet."⁽²⁴⁾

"Moving from the cost of food for a family to the total income required [i.e., the poverty threshold] entailed three basic steps. First...it was necessary to define the family size and composition prototypes for which food costs would be computed. It was then necessary to decide how much additional income to allow for items other than food, and finally how to relate the cash needs of farm families to those of their comparable nonfarm cousins. In view of the special interest in the economic status of families with children, and because logic suggests that income requirements are related to the number in the family, estimates were made separately for nonfarm families varying in size from two members to seven or more, further classified by sex of head and number of related children under 18."⁽²⁵⁾ (The reference to number of children means that among three-person families, for instance, there were separate subcategories with the following compositions: three adults; two adults, one child; and one adult, two children.) "To allow for the special interest in the aged, the majority of whom live alone or in couples, 2-person families were further classified by age of head as those under age 65 or aged 65 and older...."⁽²⁶⁾ (The fact that there are lower poverty thresholds for aged units of the smallest size is simply a mechanical consequence of the fact that there were separate subcategories for two-person families with aged and nonaged heads, and the food plan costs calculated for the aged families were lower than those for the nonaged families. Orshansky did not claim that necessary nonfood expenditures for the aged are or should be lower than those for the nonaged.) As noted below, poverty thresholds for unrelated individuals (one-person units) were not derived from food plan costs; instead, they were calculated directly from the thresholds for two-person families.

Calculating food plan costs for each of the 58 nonfarm family subcategories that Orshansky had established was a complicated process. The food plans that she was using included separate food cost figures for nineteen different age-sex categories of persons.⁽²⁷⁾ However, to a great extent data were not available to show the distribution of persons by age and sex within each of her family subcategories. She used data distributions from the 1960 Decennial Census and made additional assumptions about characteristics of family members not shown in the Census data distributions. For each subcategory of families with children, a combination of ages of children was chosen that yielded a food cost that was higher than the food costs of two thirds of the (simulated) families in that subcategory. "Because food requirements for children increase rapidly with advancing age and the food plan cost is already critically low, this protection was deemed necessary to ensure adequate allowance for growing youngsters."⁽²⁸⁾ Food plan costs for the 58 nonfarm family subcategories were calculated using January 1964 prices for the economy and low-cost food plans.⁽²⁹⁾

To get from food plan costs to estimates of minimum necessary expenditures for all items, "...an adaptation was made of a principle most of us learn by heart:"⁽³⁰⁾ "...for individuals as for nations...the proportion of income allocated to the 'necessaries,' and in particular to food, is an indicator of economic well-being."⁽³¹⁾ "...a low percentage of income going for food can be equated with prosperity and a high percentage with privation."⁽³²⁾ (The "principle" that Orshansky referred to is known as Engel's Law.⁽³³⁾,⁽³⁴⁾) Orshansky adapted Engel's Law by assuming (for families of three or more persons) that "equivalent levels of adequacy were reached only when the proportion of income required to purchase an adequate diet was identical."⁽³⁵⁾

To determine the proportion of total income that should be assumed to be spent for food, Orshansky made use of another Agriculture Department product--the Household Food Consumption Survey, a survey which the Department conducts on a recurring basis at approximately ten-year intervals. At the time she was developing the thresholds, the most recent of these surveys which was available was the one for 1955. She made use of one major finding from this 1955 survey--that for families of three or more persons, the average dollar value of all food used during a week (both at home and away from home) accounted for about one third of their total money income after taxes.⁽³⁶⁾,⁽³⁷⁾ (It should be noted that this finding relates to families at all income levels, not just families at low income levels; one of the most common errors made in describing the development of the thresholds is to assert that they are based on a finding that "poor people spend a third of their income on food." The reason for using the proportion for all families can be gleaned from the discussion on p. 6 below.)

Besides considering the Agriculture Department's 1955 Household Food Consumption Survey, Orshansky also looked at the Bureau of Labor Statistics' 1960-1961 Consumer Expenditure Survey, which also provided an estimate of the proportion of total after-tax income going for food. However, "questions employed by the Bureau of Labor Statistics to obtain the data on annual food outlays usually have yielded lower average expenditures than the more detailed item-by-item checklist of foods used in a week that serves as a questionnaire for the Agriculture Department."⁽³⁸⁾ She briefly discussed the possibility of using the BLS survey to derive a poverty measure; this would have resulted in a "multiplier" (see below) of just over four, rather than three. However, she finally decided to use the 1955 Agriculture Department survey, with its one-to-three ratio of food expenditures to after-tax money income, in developing the poverty thresholds.⁽³⁹⁾

In effect, Orshansky started her food-costs-to-total- expenditures procedure by considering a hypothetical average (middle-income) family, spending one third of its income on food, which was faced with a need to cut back on its expenditures.⁽⁴⁰⁾ She made the assumption that the family would be able to cut back its food expenditures and its nonfood expenditures by the same proportion. This assumption was, of course, a simplifying assumption or first approximation, as she herself recognized. However, she had no data to support a specific different relationship between food and nonfood expenditure cutbacks.⁽⁴¹⁾ Under this assumption, one third of the family's expenditures would be for food no matter how far it had cut back on its total expenditures.

When the hypothetical family cut back its food expenditures to the point where they equalled the cost of the economy food plan (or the low-cost food plan, in the case of her second set of thresholds) for a family of that size and composition, the family would have reached the point at which its food expenditures were minimal but adequate, assuming that "the housewife will be a careful shopper, a skillful cook, and a good manager who will prepare all the family's meals at home."⁽⁴²⁾ Orshansky made the assumption that at that point, the family's nonfood expenditures would also be minimal but adequate, and established that level of total expenditures as the poverty threshold for a family of that size. Since the family's food expenditures were still (by her earlier assumption) one third of its total expenditures, this meant that (for families of three or more persons) the poverty threshold for a family of a particular size and composition was set at three times the cost of the economy food plan (or the low- cost food plan) for such a family. The factor of three by which the food plan cost was multiplied became known as the "multiplier."⁽⁴³⁾

It is important to note that Orshansky's "multiplier" methodology for deriving the thresholds was normative, not empirical--that is, it was based on a normative⁽⁴⁴⁾ assumption involving (1955) consumption patterns of the population as a whole, and not on the empirical consumption behavior of lower-income groups.

Orshansky used somewhat different procedures to calculate poverty thresholds for one-person and two-person units⁽⁴⁵⁾, noting that "it is generally acknowledged that a straight per capita income measure does not allow for the relatively larger fixed costs that small households face. Moreover, the more recent consumption curves themselves indicate that the 1- or 2-person families, who as a group are less homogeneous in composition, seem to be 'out of line' with larger families with respect to the spending pattern."⁽⁴⁶⁾ For two-person families, the 1955 Household Food Consumption Survey had found a food/total-after-tax-money- income ratio of 0.27 rather than a third; accordingly, she used a multiplier of 1/0.27, or 3.7, to derive poverty thresholds for two-person families.⁽⁴⁷⁾

To derive poverty thresholds for unrelated individuals (one- person units), Orshansky did not use a multiplier at all; she noted that "the consumption data [for this group] are hard to interpret because of the heavy representation of aged individuals not shown separately...."⁽⁴⁸⁾ In this case, she followed slightly different procedures for her two sets of thresholds--those based (for families) on the low-cost food plan and those based on the economy food plan. For poverty "at the low-cost level," she set thresholds for unrelated individuals at 72 percent of the corresponding thresholds for two-person families, "following BLS recent practice."⁽⁴⁹⁾ For poverty at "the economy level"--that is, the definition of poverty that is still in use today--she set the thresholds for unrelated individuals at 80 percent of the corresponding thresholds for two-person families, "on the premise that the lower the income[,] the more difficult it would be for one person to cut expenses such as housing and utilities below the minimum for a couple."⁽⁵⁰⁾ Note that Orshansky used the 80 percent factor to derive separate thresholds (at the "economy level") for male aged, male non-aged, female aged, and female non-aged unrelated individuals. However, because of the different weighting factors involved for one-person and two- person units, the weighted average poverty threshold for a one- person unit is not precisely equal to 80 percent of the weighted average poverty threshold for a two-person unit.

Having calculated poverty thresholds from each food plan for 58 categories of nonfarm families and 4 categories of nonfarm unrelated individuals, Orshansky had 62 detailed poverty thresholds (from each food plan) for nonfarm family units. Her next step was to develop separate detailed thresholds for the corresponding categories of farm family units.⁽⁵¹⁾ She noted that in 1955 (according to the most recent Household Food Consumption Survey) "about 40 percent of the food items consumed by all farm families--valued at prices paid by any families who did buy them--came from their home farm or garden [rather than being purchased for cash]....Farm families generally can count not only some of their food but most of their housing as part of the farm operation."⁽⁵²⁾ Because farm families purchased for cash only about 60 percent of the food they consumed, and because of the issue of classifying farm housing expenses as part of the farm business operation, Orshansky decided to set farm poverty thresholds at 60 percent of the corresponding nonfarm thresholds.⁽⁵³⁾ (As noted below, this figure was changed to 70 percent in 1965 when OEO adopted the thresholds, to 85 percent in 1969, and to 100 percent--that is, the differential was eliminated--in 1981.)

It is important to note that Orshansky's farm/nonfarm distinction was not the same as a rural/urban (or nonmetropolitan/metropolitan) distinction. In April 1970, for instance, the Decennial Census found that of a total rural population of 53.9 million persons, only 10.6 million (19.7 percent) lived on farms. The nonfarm poverty thresholds were applied to the rural nonfarm population as well as to the urban population. It should also be noted that the reason for the farm/nonfarm distinction was not a generalized "Living-costs-are- cheaper-in-farm-or-rural-areas" argument.

With 62 detailed poverty thresholds for nonfarm family units and 62 detailed thresholds for farm family units, Orshansky had a total of 124 detailed thresholds at each of the two cost levels (low-cost and economy). Instead of doing a tabular presentation of 248 separate income cutoff figures, she decided to present a smaller set of weighted average thresholds.⁽⁵⁴⁾ The weighted average nonfarm poverty thresholds at the economy level and at the low-cost level for calendar year 1963 are shown in the table below.⁽⁵⁵⁾

Poverty thresholds at Ratio of Poverty thresholds low-cost level low-cost Size of at economy level [near-poverty/low-income thr. to family unit [poverty thresholds] thresholds] poverty thr.

1 person $1,539 NA NA (under age 65) 1,580 $1,885 1.19 (aged 65 or over) 1,470 1,745 1.19

2 persons 1,988 NA NA (head under 65) 2,050 2,715 1.32 (head 65 or over) 1,850 2,460 1.33

3 persons 2,440 3,160 1.30 4 persons 3,130 4,005 1.28 5 persons 3,685 4,675 1.27 6 persons 4,135 5,250 1.27 7 or more persons 5,090 6,395 1.26

Note that calendar year 1963 has always been the base year for the poverty thresholds, both before and after the 1969 revision discussed below.⁽⁵⁶⁾

When Orshansky was developing the poverty thresholds, the Census Bureau's Current Population Survey (CPS) was the only good source of nationally representative income data. Accordingly, Orshansky had to apply her poverty thresholds to the CPS income data, even though the CPS used a before-tax money income concept, while the thresholds had been developed on the basis of the after-tax money income concept used in the Agriculture Department survey from which the multiplier was derived. Orshansky was aware from the beginning of the inconsistency of applying after- tax thresholds to before-tax income data, but had no other alternative; she reasoned that the result would yield "a conservative underestimate" of poverty. As she noted, at that time (as well as for some years thereafter) most families and individuals at the poverty level had little or no federal income tax liability. (Some did, however, pay Social Security payroll taxes and/or state income taxes.)⁽⁵⁷⁾

The poverty thresholds were presented as a measure of income inadequacy--in Orshansky's words, "if it is not possible to state unequivocally 'how much is enough,' it should be possible to assert with confidence how much, on an average, is too little."⁽⁵⁸⁾ In a 1965 SSA memo, it was not the poverty level but the near- poor or low-income level--the thresholds derived from the low- cost food plan--that was described as "a minimum adequate standard."⁽⁵⁹⁾

Orshansky accurately described her poverty thresholds as a "relatively absolute" measure of poverty⁽⁶⁰⁾, inasmuch as they were developed from calculations that made use of the consumption patterns (at a particular point in time) of the U.S. population as a whole. (In the dichotomy between relative and absolute definitions of poverty, one of the essential characteristics of a purely "absolute" definition of poverty is that it is derived without any reference to the consumption patterns or income levels of the population as a whole.⁽⁶¹⁾) However, while Orshansky's poverty thresholds were not a purely absolute measure, they were also quite clearly not a purely relative measure, such as the 50-percent-of-median-income measure proposed by Britain's Peter Townsend in 1962 and (in the United States) by Victor Fuchs in 1965.⁽⁶²⁾

Poverty lines--including Orshansky's--have often been called "arbitrary." The application of the term "arbitrary" to poverty lines can be traced back at least as far as 1915, when two Englishmen--a statistician and an economist--described a pair of English standard-budget-based poverty lines as "being, to a considerable extent, abstract and arbitrary."⁽⁶³⁾ Orshansky also applied the term to her poverty thresholds in her January 1965 article. However, the specific wording that she used was "arbitrary, but not unreasonable."⁽⁶⁴⁾ Subsequent writers have often repeated the first phrase while ignoring the second. An additional point worth noting about the word "arbitrary" is that it has several distinct connotations, including: 1) depending on judgment, choice, or discretion (used in particular of the decision of a judge as contrasted to a decision or sentence specified in a statute); and 2) random or capricious. Orshansky used the word with the first connotation, while a number of subsequent writers seem to use it with the second connotation.

Based on an extensive study of over forty poverty lines and subsistence budgets developed in the U.S. during the 1900-1965 period, the author's assessment is that the analysts who developed those poverty lines were not merely picking "arbitrary," capricious numbers at random. Instead, they were generally trying to develop figures that approximated a rough social consensus about the level of a socially acceptable minimum standard of living at a particular time.⁽⁶⁵⁾ This assessment applies to Orshansky's poverty thresholds also; and there is some specific evidence that her thresholds approximated a rough social consensus about an acceptable minimum standard of living during the early 1960's both among experts and among the general population:

o When one looks at twelve expert-developed poverty or low- income lines applied to families of four during the 1959- 1964 period, one finds that eight of them were between $3,000 and $3,500 in current dollars⁽⁶⁶⁾, showing a rough consensus among experts⁽⁶⁷⁾ during this period about the level of a socially acceptable minimum standard of living. Orshansky's poverty threshold of $3,128 (in 1963 dollars) for a nonfarm family of four fit in well with this expert consensus. (As Orshansky herself noted in an April 1963 memo (see p. 15 below), "A variety of criteria have been used to produce [poverty] estimates of the same order of magnitude.")

o We do not have any direct evidence about the views of the general population in 1963 about the level of a socially acceptable minimum standard of living--i.e., about where the public would have placed a poverty line at that time. However, in 1993 Denton Vaughan constructed a socially defined poverty standard for the 1947-1989 period using Gallup Poll responses to the "get-along" question for those years and a Gallup Poll response to a question specifically about the poverty line in 1989.⁽⁶⁸⁾ (He assumed that the ratio of the poverty line response and the "get-along" amount for 1989 could be applied to the earlier years.) His Gallup-Poll-based poverty standard for a family of four for 1963 was $3,108--almost identical to Orshansky's 1963 poverty threshold of $3,128 for a nonfarm family of four. In Vaughan's words (p. 28), this suggests that Orshansky's poverty line "was generally consistent with societal notions about the poverty level prevailing at about the time it was introduced."

Orshansky's landmark work of developing her poverty thresholds is well known not only in the United States but also among poverty researchers in such countries as Canada, Great Britain, the Netherlands, and Australia.⁽⁶⁹⁾

The Historical Background and the Early History of the Poverty Thresholds

Concerning the genesis of the poverty thresholds, Orshansky wrote the following in a 1988 introduction to a reprinting of her January 1965 article:

The Social Security Administration (SSA), in carrying out its basic research mission, regularly assesses and reports on the economic well-being of selected groups. Widows under age 62 with minor children was one such group for whom an annual income series was developed to compare the economic status of these "young survivors" and of other women bringing up young children without a father present in the household. By 1962, changing health and industrial conditions resulted in too few new "orphans" each year to warrant an annual report. However, the number of children with no father in the household for reasons other than death was growing and their family groups had a less favorable income status than the young survivors. In an exploratory mode, 1961 income data for all families with children were requested from the March 1962 CPS [Current Population survey]. The results were striking: As a group, mother- child families averaged less than half the income of two- parent families, but many of the latter had low income also. To suggest the insufficiency of family funds for the rearing of children, "crude indexes" of poverty at two levels were constructed by relating minimal food costs to family income. The findings were published in the July 1963 Social Security Bulletin article "Children of the Poor."...Meanwhile, the Council of Economic Advisors planning for the War on Poverty to be proclaimed in 1964 was using $3,000 as the poverty line for a family of two or more....The SSA research plan to extend the crude index [of poverty] to families without children was accelerated and ["Counting the Poor: Another Look at the Poverty Profile"] resulted, followed by a number of analyses for subsequent years.⁽⁷⁰⁾

In a 1971 interview, Orshansky had given a little additional information about the genesis of the thresholds:

In 1963, I wrote a paper called..."Children of the Poor." It grew from a research project I was doing which was part of social security business. My boss, Mrs. Merriam [Ida Merriam, Assistant Commissioner for Research and Statistics], had asked for a general study on poverty among different types of families, not just the aged. She saw the potential uses for the information that would be gained....I picked children for my research because personally I happen to like children better than people....Since I had to indicate which kids I was talking about, I developed a crude statistical measure of poverty to work with.⁽⁷¹⁾

A 1970 interview of Orshansky noted that her "years at [the U.S. Department of] Agriculture were invaluable training for her future development of the 'poverty line,' since she was involved with living standards and spending patterns of American farm families." She also became "[f]amiliar with the sample diets [food plans] for upper, middle and low-income families devised by the Agriculture Dep[artmen]t," later using one of these Agriculture Department food plans "as the basis of her poverty yardstick."⁽⁷²⁾ In a 1989 interview, Orshansky noted, "...I had worked in [the Department of] Agriculture, in what was at that point the Bureau of [Human Nutrition and] Home Economics...what happened was that I got to know some of the things that 'the ladies' were interested in. I knew what they were doing on food and nutrition."⁽⁷³⁾ "[W]hat they were doing on food and nutrition" included not only the Agriculture Department food plans but also the 1955 Household Food Consumption Survey, which Orshansky also used in developing the poverty thresholds.

Although not mentioned by Orshansky in the above article and interviews, direct antecedents of her poverty thresholds go back several years before "Children of the Poor" in 1963. On April 11, 1960, Arthur S. Flemming, Secretary of Health, Education, and Welfare, was testifying before a Senate subcommittee on the subject of health needs of the aged and aging. Senator Pat McNamara [D-Mich.], the subcommittee chairman, asked him, "Do you have any figures on how much it costs a retired couple to live these days? What are your recent figures?" Secretary Flemming said that he did not have any such figures with him at the time, but "We would be very glad, in cooperation with the Department of Labor, to develop some figures and submit them for the record at this point."⁽⁷⁴⁾ For some reason, the Labor Department did not provide material to respond to the senator's question, so Orshansky prepared a response, which was submitted by HEW (without attribution) and printed as "Income Needs of the Aged."⁽⁷⁵⁾ In this response, she mentioned the SSA's 1948 budget for an elderly couple, which had not been updated since 1950, as well as a similar current budget for elderly people in New York City, with October 1959 cost figures for a couple and for male and female individuals living alone. She noted that the Bureau of Labor Statistics was in the process of revising the budget for an elderly couple. (As noted in footnote 15, she wrote an article about this BLS revision of the budget later in the same year.) She went on to state that
Other means of approximating income need can be devised. For example, the total can be estimated from the cost of food, which is one category for which generally accepted standards of adequacy are available. [She then gave the January 1960 cost of the Agriculture Department's low-cost food plan--at that point the cheapest of three food plans-- for an elderly couple.] Analyses from the most recent nationwide food consumption study carried out by the U.S. Department of Agriculture, in spring 1955, suggest that families with a homemaker age [sic] 60 or older, were spending for food in a week at a rate equivalent to only about one-fourth (27 percent) of the year's money income after taxes, compared with one-third for families of all ages. On the premise, therefore, that food should not claim more than one-fourth of the income of an aged couple, it can be said on the basis of the low-cost food plan that an income less than $2,560 for an elderly couple is probably "uncomfortably low." This is just about the same as the income ($2,675) exempted from Federal income tax for a couple with both members 65 and over, who take no more than a standard deduction. By the same token, an income of less than $2,080 for the couple might be termed inadequate, because at current prices more than $1 out of $3 would have to be devoted to food to get an adequate diet at low cost.

In other words, Orshansky used an Agriculture Department food plan and results from the Agriculture Department's 1955 Household Food Consumption Survey to derive two measures of low or inadequate income--almost exactly as she was to do several years later in her July 1963 and January 1965 articles.

About eight months later--on December 19, 1960--the office⁽⁷⁶⁾ in which Orshansky worked issued a Research and Statistics Note, "Facts About Financial Resources of the Aged," which was specifically credited to her.⁽⁷⁷⁾ This note was "[p]repared...for the Chartbook for the White House Conference on Aging, January 9- 12, 1961." On p. 2 of the note, after mentioning the just- released costs of the BLS revised budget for an elderly couple, Orshansky went on to quote almost verbatim from the material on "Income Needs of the Aged" that she had prepared earlier in the year:

...on the premise--consistent with the spending patterns of older families--that food should claim no more than one- fourth of income, it might be said, based on the Low-Cost Food plan of the U.S. Department of Agriculture, that income less than $2400-$2,500 [sic] for an elderly couple is uncomfortably low.

This sentence was not carried over into the actual Chartbook for the White House Conference on Aging.

These two 1960 items of Orshansky's were not Social Security Bulletin articles, and they contained no analysis of the population below the income cutoffs described. Yet they did contain the essence of the poverty threshold concept presented in her July 1963 and January 1965 articles. However, with only one exception⁽⁷⁸⁾, the writer has never seen the two 1960 items cited or referred to in print. The complete contrast between the fate of these 1960 items and the fate of Orshansky's 1963 and 1965 work is an outstanding illustration of her own comment, "Apparently the right timing is as important as the right idea."⁽⁷⁹⁾ In 1960, most middle-class Americans and most social scientists were just not giving any thought to the subject of poverty in America.⁽⁸⁰⁾

As Orshansky noted (see p. 11 above), her work on what became "Children of the Poor" came in the wake of the termination of the series of reports on the incomes of "young survivors" (widows under age 62 with minor children). The articles that she did on this group were published in September 1959, September 1960, and October 1961 (see footnote 70). If another article in that series had been published, it would probably have come out in September or October 1962. Accordingly, the decision to discontinue the series of reports may have come as late as the summer of 1962. As will be seen below, the decision to involve Orshansky in what became "Children of the Poor" may have come in or shortly after January 1963.

Parts of the history of the project that became "Children of the Poor" can be traced in several documents found in the [Ida] Merriam Collection of Social Security Papers now located in the Gelman Library of George Washington University in Washington, D.C. A Fiscal Year 1963 work plan⁽⁸¹⁾ for the Division of Program Research--apparently but not certainly prepared before the July 1962 beginning of that fiscal year--contained (p. 7) a brief description of Project DPR5, Poverty as it Affects Children: "Analyses of the financial resources of families with children, and of the general circumstances in which these families live. A monograph will be prepared including estimates of number of children living in poverty, and the impact of social security and social welfare programs on normal [sic] families and on families broken by death or marital discord." An estimated completion date of December 1964 was given, with the comment "(1/4 completed in fiscal 1963)"--apparently, from the context, meaning that one quarter of the project was to be completed in Fiscal Year 1963 (the year ending June 30, 1963). This project was included together with at least four others under the heading "Family Economics Studies."

An October 3, 1962, memorandum (from Ida C. Merriam, Director, Division of Program Research, to Robert M. Ball, Commissioner of Social Security--Subject: Studies of poverty among families with children) mentions "a number of special and more limited studies we could do in the course of the next year." For one of these special studies, "The priority might well be raised on Project DPR 5 in our F[iscal] Y[ear] 1963 Work Plan (Poverty as it Affects Children)." The memo ends, "Will you let me know your general reaction to the idea of our markedly stepping-up our research relating to poverty among families with children as part of the SSA contribution to current understanding of the situation in AFDC [Aid to Families with Dependent Children]." (The reference is to a number of projects--including but not limited to Project DPR 5.) A handwritten response from "RMB" [Robert M. Ball] to "Ida" at the bottom of the first page of the memo says, "I am all for it. I think a concentration on problems of children in low income families with many projects over next several years is just the right emphasis."

A memorandum (Revised January 3, 1963, from Robert A. Dentler to Ida C. Merriam--Subject: Research Priorities in the Social Security Research Institute) does not mention Project DPR 5 as such, but does say "the Institute should attend to the urgently needed study of selected subgroups of household units defined as least adequate" (p. 7--emphasis in original). On the back of this 17-page memo, handwritten notes include the notation "selected subgroups of least adequate households[--]Alvin Schorr or Mollie". (Alvin Schorr was Mollie Orshansky's immediate supervisor--personal communications with Mollie Orshansky, June 14, 1988, and February 10, 1989.)

The above documents show a clear interest in SSA in 1962 and 1963 in the problems of economically insecure households, especially poor families with children; they fit in well with Orshansky's statement that "'Children of the Poor'....grew from a research project I was doing which was part of social security business. My boss, Mrs. Merriam, had asked for a general study on poverty among different types of families, not just the aged" (see pp. 11-12 above).

The handwritten notation on the back of the January 1963 memo strongly suggests that Orshansky became involved in a study of certain "least adequate households" in or shortly after January 1963. By April 11, 1963, she had completed a 14-page memorandum to Ida C. Merriam called "The Children of the Poor - Some Suggestions for Research"; this memo contained much of the material that was included two months later in "Children of the Poor," as well as a number of proposals for research projects. In discussing the limitations of the "crude" poverty measures that she was putting forward, Orshansky commented,

Identifying the deserving poor is presently almost a national pastime--although doing something about them has not yet achieved such popularity. A variety of criteria have been used to produce estimates of the same order of magnitude. The similarities in result are striking enough to engender a feeling either of helpless disquietude, or defensive disbelief. This is, after all, an enlightened economy and the thought of hungry children is unpleasant.

She discussed her proposals for research under the headings "Assay of existing programs," "Wage mobility," "The legacy of poverty," "Beneficiary studies," "Studies of educational opportunities," "Living with poverty," "Defining poverty," and "International comparative studies." Under "The legacy of poverty," she wrote, "A special sample of families with children, oversampled for those with low income, should be designed to yield enough broken families and low-income husband-wife families of whom one might ask retrospective questions. (A prospective study with follow-up interviews at specified intervals for a period of 5 to 10 years would be more useful....)...."⁽⁸²⁾ She concluded a section on "Implications for research" with the words "We must continue to look for the basic causes [of poverty] but we cannot wait for success before we treat the symptoms. We may find as often happens in medicine that learning what measures are successful (or unsuccessful) in bringing relief helps narrow the list of suspected etiologic agents."

In July 1963, Orshansky's article "Children of the Poor" was published in the Social Security Bulletin. Much of the analysis in the article was based on a special tabulation of data from the March 1962 Current Population Survey (covering incomes for calendar year 1961) which SSA purchased from the Census Bureau.⁽⁸³⁾ "The cost of...the tabulation, [Orshansky] recalls, was $2,500 and the results showed that the median income of a mother [female family head] with children was [about] $2,300 a year. 'I got sick about the fact that what the government [SSA] paid for one tabulation was more than what half of these families had to live on for an entire year,' Miss Orshansky reported. "I determined I was going to get my $2,500 worth.' She did just that--and then some."⁽⁸⁴⁾

The opening paragraphs of the article contained two statements reflecting the concept now known as the income elasticity of the poverty line: "Creature comforts once the hallmark of luxury have descended to the realm of the commonplace, and the marvels of modern industry find their way into the home of the American worker as well as that of his boss....As the general level of living moves upward and expands beyond necessities, the standards of what constitutes an irreducible minimum also change."⁽⁸⁵⁾ The actual discussion of Orshansky's "standards of...an irreducible minimum" in the eleven-page article was relatively brief, being confined to two pages. It began, "A crude criterion of income adequacy--that the low-cost food plan priced by the Department of Agriculture in January 1962 represents no more than one-third of total income-- consigns about 71 percent of the mother-child families to low- income status [in calendar year 1961]. Even the use of the [Agriculture] Department's economy [food] plan, estimated to cost about 20 percent less than the low-cost plan, leaves at 61 percent the proportion of the mother-child families who must devote to food more than $1 out of $3 to get a nutritious diet."⁽⁸⁶⁾ The relevant table designated families and children as being "Poor by low-cost diet" and "Poor by economy diet."⁽⁸⁷⁾ Orshansky did not present a table of poverty lines for families, but did end the discussion by stating, "By way of suggesting the level of living implied by the present approximation, the income required for a husband, wife, and two children not on a farm would be $3,165 by the more conservative [economy] measure, or $3,955 by the more liberal [low-cost measure]. The mother-and- two-child family, with allowance for the additional relative assumed to be living with the family, would require $2,945 or $3,680."⁽⁸⁸⁾

As can be seen, Orshansky presented the low-cost-food-plan- based poverty thresholds first in her July 1963 article. Besides presenting the economy-food-plan-based thresholds second, she introduced them with the dismissive adverb "Even." Similarly, in her table presenting poverty figures, the column heading "Poor by low-cost diet" came before "Poor by economy diet." These facts suggest that the thresholds based on the low-cost food plan may well have been the poverty measure that she preferred. After the original version of the present paper was completed, Orshansky confirmed this inference (personal communication, August 23, 1994), stating that the thresholds based on the low-cost food plan were indeed her preferred version of the poverty measure. (Similar sentiments can be detected "under the surface" in comments that she made in a 1969 article: "At the Social Security Administration, we decided that we would develop two measures of need....It was not the Social Security Administration that labeled [one of these] the poverty line. It remained for the Office of Economic Opportunity and the Council of Economic Advisers to select the lower of the two measures and decide they would use it as the working tool....It is interesting that few outside the Social Security Administration ever wanted to talk about the higher measure."⁽⁸⁹⁾ And note that in a May 1964 memorandum (see pp. 20-21 below), Orshansky described the low- cost-food-plan-based poverty measure as "probably more realistic" than the economy-food-plan-based poverty measure.)

In his January 1964 State of the Union address, President Lyndon Johnson announced a War on Poverty. At the President's express wish⁽⁹⁰⁾, the 1964 Report of the Council of Economic Advisers (CEA) contained a chapter on "The Problem of Poverty in America."⁽⁹¹⁾ The chapter set a poverty line of $3,000 (in 1962 dollars) for families of all sizes; for unrelated individuals, the chapter implicitly set a poverty line of $1,500 (a selection which was shortly made explicit). "Refined analysis would vary the income cut-off by family size, age, location, and other indicators of needs and costs. This has not been possible." The $3,000 figure was specified as being on the basis of before-tax annual money income. There was a brief discussion of the theoretical desirability of using estimates of "total incomes-- including nonmoney elements....such...as the rental value of owner-occupied dwellings and food raised and consumed on farms...", but it was not possible to obtain such estimates. "Of course, the total of money plus nonmoney income that would correspond to the limit [the poverty line of $3,000 in money income] used here would be somewhat higher than $3,000."⁽⁹²⁾

After referring to the poverty lines in "Children of the Poor" (see next paragraph) and their assumption of one third of money income going for food, the chapter indicated that out of its $3,000 poverty "budget," $1,000 could be assumed to go for food. "...a conservative estimate for housing (rent or mortgage payments, utilities, and heat) would be another $800."⁽⁹³⁾ The $800 figure was presumably loosely based on the contemporary "rule of thumb" that one fourth of a family's income "ought" to go for housing; this figure will have come from the CEA staff, since the Orshansky poverty lines did not contain any assumption whatsoever about what proportion of a poverty income ought to go for housing.

The chapter began its discussion of a poverty line by stating that

...society does not have a clear and unvarying concept of an acceptable minimum [income standard]....But for our society today a consensus on an approximate standard can be found. One such standard is suggested by a recent study, described in a publication of the Social Security Administration, which defines a 'low-cost' budget for a nonfarm family of four...[as] $3,955. The cost of what the study defined as an 'economy-plan' budget was $3,165. Other studies have used different market baskets, many of them costing more. On balance, they provide support for using as a boundary, a family whose annual money income from all sources was $3,000 (before taxes and expressed in 1962 prices).⁽⁹⁴⁾

People who have read this passage generally think that it means that the CEA's $3,000 poverty line was derived to a greater or lesser degree from Orshansky's $3,165 (economy-food-plan-based) poverty line. However, several printed sources⁽⁹⁵⁾ note that Robert Lampman (a member of the CEA staff) had been working on an analysis of poverty using the $3,000 figure as early as the spring of 1963, providing data to CEA Chairman Walter Heller for a memorandum to President Kennedy dated May 1, 1963. Since "Children of the Poor" was not published until July 1963, it could not have determined the choice of the $3,000 figure, despite its being referred to in the January 1964 CEA report chapter in connection with the $3,000 figure.

Lampman (personal communications, September 10, 1987, and October 30, 1993) provided the following specific information about the development of the CEA's $3,000 poverty line in a 1987 telephone conversation and a 1993 interview with the author. (Lampman was the primary author of the CEA report's poverty chapter.⁽⁹⁶⁾ He had been brought on to the CEA staff because of his 1959 paper on the low income population.⁽⁹⁷⁾)

o Orshansky's $3,165 poverty figure from "Children of the Poor" was not used in deriving the CEA's $3,000 poverty line.

o The poverty-as-half-of-median-income concept was not a consideration in setting the $3,000 poverty line. Lampman specifically denied later claims that the $3,000 figure was an attempt to covertly introduce a half-of-median-income poverty line.⁽⁹⁸⁾

o There was no connection between the CEA's $3,000/$1,500 poverty line and the AFL-CIO's 1960 poverty line⁽⁹⁹⁾ using the same figures; indeed, Lampman had not been aware of the AFL- CIO poverty line.

o The CEA's $3,000 poverty line was a "consensus" figure based on several separate considerations. It was roughly the amount that someone would earn if working year-round at the minimum wage.⁽¹⁰⁰⁾ It was at the approximate level at which a family [of four] started paying taxes.⁽¹⁰¹⁾ It was not too far above the highest state payment under the Aid to Families with Dependent Children program.⁽¹⁰²⁾

o Gardiner Ackley [a member of the Council of Economic Advisers] decided that the CEA's poverty line for the 1964 Economic Report should not vary by family size. He said that a poverty line with different figures by family size would be too complicated for this sort of document--that a single number would be enough. He was, however, willing to have the refinement of an adjustment for family size introduced later.

It appears that Orshansky was aware that her $3,165 figure had not actually been used in deriving the CEA $3,000 figure.⁽¹⁰³⁾

When Orshansky saw the January 1964 CEA report, with its reference to her July 1963 work, she was disturbed by the CEA's failure to vary its $3,000 family poverty line by family size.⁽¹⁰⁴⁾ The CEA "standard led to the odd result that an elderly couple with $2,900 income...would be considered poor, but a family with a husband, wife, and four little children with $3,100 income would not be."⁽¹⁰⁵⁾ "Inevitably this led to an understatement of the number of children in poverty relative to aged persons. And it was...this inequity"⁽¹⁰⁶⁾ that concerned Orshansky. Someone raised the possibility of her finishing the work on poverty thresholds that she had started in "Children of the Poor." She was called in one Saturday and asked if she could extend her families-with-children poverty lines to the whole population. She said that she could--but that she couldn't complete that assignment in only three or four weeks, no matter how much funding they gave her to do it. It was decided that she and the Division of Research and Statistics would proceed to extend her poverty lines to the rest of the population.⁽¹⁰⁷⁾

As an early step in this process, Lenore Epstein issued a Research and Statistics Note⁽¹⁰⁸⁾ in late February 1964 which included poverty lines for an elderly couple derived from the low-cost and economy food plans by applying the multiplier procedure that Orshansky had described in her April 1960 "Income Needs of the Aged" (see footnote 75). Perhaps at the same time-- and definitely by March 19--a poverty line for an unrelated individual based on the economy food plan had also been calculated.⁽¹⁰⁹⁾ In early April, Ida Merriam wrote,

For a more exact measure of poverty [than the CEA's $3,000 figure for families] what is needed is a series of income cut-off points that represent equivalent levels of living for families of different size and type. Just as there is no one standard of minimum adequacy, so there is no definitive scale of equivalence....The Division of Research and Statistics is planning to buy from the Bureau of Labor Statistics special tabulations from the 1960-61 [consumer] expenditure survey that we think will give us a better basis for developing income-consumption equivalence scales than anything now available. This work will be time consuming. It may well be [the] fall of 1965 before a new scale can be developed. In the meantime, we are calculating a rough scale based on the Department of Agriculture economy-cost food budgets for families of different types and the average relationship of food to income shown by earlier consumption studies for several types of families....We are now discussing with the Census Bureau the possibility of getting runs of Census [Current Population Survey] data for 1963 showing the number of persons in families with incomes below the poverty level, as calculated on this economy cost food standard, for about 35 different family types. Neither the cost nor the time required is definite as yet.⁽¹¹⁰⁾

In late May, Orshansky wrote a memorandum to the Census Bureau enclosing

an experimental set of poverty-line criteria for farm and nonfarm families....[derived from] the Agriculture Department's economy and low-cost food budgets....Following our earlier discussions, it is my understanding that these income tests are to be applied to the data obtained in the March 1963 Current Population Survey.⁽¹¹¹⁾

The poverty lines enclosed were differentiated by family size, number of family members who were children, and age (for one- and two-person units only), but not by sex of family head. While this memo discussed applying the figures to March 1963 CPS data, March 1964 CPS data (covering incomes for calendar year 1963) were used for the final version of Orshansky's January 1965 article. The enclosure to the memo described the economy-food- plan-based poverty measure as "a most conservative measure," and the low-cost-food-plan-based poverty measure as "a more generous but still conservative measure. It is probably more realistic."

Orshansky had completed the work for "Counting the Poor: Another Look at the Poverty Profile" by late 1964⁽¹¹²⁾, and the article was published in January 1965. The subtitle of the article may imply a contrast with the CEA's January 1964 chapter as a first look at the poverty profile. Much of the substance of the article has been discussed above in connection with the development of the thresholds.

In the opening pages of the article, Orshansky first mentioned her poverty lines based on the economy food plan, and then mentioned the "somewhat less conservative but by no means generous standard" based on the low-cost food plan.⁽¹¹³⁾ (This reversed the order of presentation in "Children of the Poor.") She generally distinguished the two sets of poverty lines by means of the phrases "the economy level" and "the low-cost level." When she used the term "poor" without further qualification, she was generally referring to poverty at the economy level.

As noted, preprints of Orshansky's January 1965 article "were circulated fairly widely before their formal publication..." (see footnote 112). The CEA's 1965 report, also published in January, contained two paragraphs on "Differences in family composition" noting that SSA had developed poverty lines varying by family size and other factors, and that these poverty lines [at the economy level], while not changing the poverty population total very much from that under the CEA's January 1964 definition, did result in fewer aged persons and many more children in the poverty count.⁽¹¹⁴⁾ Preprints of Orshansky's article "land[ed] in enthusiastic laps just at the time Sargent Shriver was beginning to implement the [Economic Opportunity] Act of 1964 [by setting up the Office of Economic Opportunity]. Hy[man] Bookbinder brought a copy [of the article] from the Council of Economic Advisers to Shriver [the Director of OEO]. The latter referred it to Leon Gilgoff, Acting Director of the Office of [Research, Plans, Programs,] and Evaluation, with marginal notes recommending serious consideration for official adoption by OEO."⁽¹¹⁵⁾ On March 7, 1965, Gilgoff sent a briefing memorandum on the Orshansky poverty lines to Shriver, describing them as a "second generation definition of poverty"⁽¹¹⁶⁾--meaning that they represented a significant step beyond the CEA's initial $3,000/$1,500 poverty line. On March 12, Joseph Kershaw of OEO's Office of Research, Plans, Programs, and Evaluation gave a presentation before the Economic Opportunity Council [a council whose official members included a number of Cabinet members] which included material on the "Second Generation Definition of Poverty." One chart in this briefing, titled "Economy Level vs. Low-Cost Level," compared the poverty population, the poverty rate, and the poverty gap under the two definitions of poverty, but the other four poverty charts all used the economy level poverty definition.⁽¹¹⁷⁾ The new poverty measure (particularly the one at the economy level) seems to have met a favorable reception, although it was not officially adopted by OEO at this time.⁽¹¹⁸⁾

On April 5, 1965, SSA convened a meeting of an ad hoc interdepartmental advisory group "to advise regarding several technical questions immediately at issue in preparing final specifications for SSA's special tabulations on income from the March 1965 Supplement to the Current Population Survey."

There was general agreement that conceptually the poverty cut-off points must be adjusted for price change, even when the change is small....Faith Clark [Director, Consumer and Food Economics Research Division, Agricultural Research Service, Department of Agriculture] suggested that a better measure [than the Consumer Price Index] of price changes for this purpose would be the change in the per capita cost of the economy food budget (which involves a different weighting of foods than the overall price index). It was agreed that this was a technically better procedure [and it was adopted].⁽¹¹⁹⁾

(Specifically, it was the December-to-December change⁽¹²⁰⁾ in the per capita cost of the economy food plan that was used to update the thresholds for annual price changes.)

At the same meeting,

The group gave particular attention to the farm-nonfarm relationship in the index. After considering such points as the relative presence or absence of public services, the relative cost of housing for farm and nonfarm families, and the generally recognized understatement of income of farm families in the Census [Current Population Survey] data, the group agreed that the basic methodology used by SSA to establish a relationship was the only method consistent with the underlying approach. It was also agreed, however, that the data on the proportion of home produced food from the new [1965] Department of Agriculture [Household Food Consumption] survey should be substituted for the 60 percent derived from the 1955 survey. (At the time of the meeting, the final figures were not available. They have since been provided by the Department of Agriculture and as a result we have moved the ratio up to 70 percent.)⁽¹²¹⁾

(The specific decision to use a 70 percent farm/nonfarm differential in the SSA poverty tabulations appears to have been made at a meeting in May⁽¹²²⁾; presumably this is when the relevant data from the new Agriculture Department survey became available.)

OEO adopted the lower (economy level) of Orshansky's two sets of poverty thresholds as a working definition of poverty for statistical, planning, and budget purposes in May 1965⁽¹²³⁾; this adoption was announced by Sargent Shriver, Director of OEO, on May 2.⁽¹²⁴⁾ From this point until 1969 (see below), the Orshansky poverty thresholds were the quasi-official federal definition of poverty.⁽¹²⁵⁾ A May 10 internal OEO briefing memorandum on the new poverty definition noted that while the poverty thresholds at the low-cost level "cannot be characterized as excessive," the thresholds at the economy level had been selected as OEO's poverty definition "on the premise that the first order task of the War Against Poverty is to get at the hard-core poor."⁽¹²⁶⁾

In her January 1965 article, Orshansky had avoided repetition of a single term by using a number of synonyms-- "poverty cutoff points," "poverty line," "income cutoffs," "standards," and "poverty income criteria." She did not, however, use the term "poverty thresholds" in this article. Her first use of the term "poverty threshold" which the writer has found was in a paper presented on May 24, 1965, at the annual meeting of the National Conference on Social Welfare.⁽¹²⁷⁾ In her later articles, she added the new term to the other synonyms that she used for the same concept.

In late 1965, the CEA asked SSA to provide trend data on poverty in years before 1963 for the CEA's 1966 report.(128) SSA provided prior-year data for 1959-1962, as well as the already available data for 1963 and 1964. These data were published in the CEA's 1966 report⁽¹²⁹⁾ and in an April 1966 Social Security Bulletin article.⁽¹³⁰⁾ Combining information from written sources and conversations with Orshansky, it appears that Current Population Survey tapes were available as far back as income year 1958, but the tabulations were begun with income year 1959 because of a change in the definition of "farm" which became effective at that point.⁽¹³¹⁾

All of Orshansky's articles and analyses of poverty used tabulations of Census Bureau (Current Population Survey) data. The first appearance of statistics using her poverty concept in a Census Bureau publication was in March 1966, in an advance report of results of an OEO-financed survey of Watts and adjoining areas of Los Angeles in the wake of the riots there during the summer of 1965.⁽¹³²⁾ The first appearance of national-level poverty population data in a Census publication was in a table in an August 1967 advance report on 1966 family income⁽¹³³⁾, while the first full Census Bureau report on the subject of poverty was issued in May 1968.⁽¹³⁴⁾

In "Recounting the Poor...", her April 1966 Social Security Bulletin article, Orshansky indicated (p. 20) that her lower poverty measure--the one at the economy level--had "now generally [been] adopted as the poverty level...", while her "somewhat less stringent measure [at the low-cost level] has now been designated as the 'near poor' level. Persons rated poor or near poor by these measures can be said to be in the 'low-income' category." On the same page, she also referred to the near poor level as "the low-income threshold."

The 1969 Revision of the Poverty Thresholds

Beginning less than twelve months after Orshansky's poverty thresholds at the economy level had been adopted by OEO as the federal government's quasi-official definition of poverty, SSA personnel began to express concern about how the thresholds should be affected by the historical fact that poverty/subsistence measures have tended to rise in real terms as the real incomes of the general population have risen.⁽¹³⁵⁾ (Orshansky had in effect set the scene for this concern by her comment in her July 1963 article (p. 3), "As the general level of living moves upward and expands beyond necessities, the standards of what constitutes an irreducible minimum also change.")

o In the November 1965 memo cited in footnote 128, Robert M. Ball (the head of SSA) wrote (p. 3), "Measures of income adequacy (or of poverty) change over time with the rise in general levels of living....one of the most difficult methodological questions we will have to face in the next few years is when and how to adjust the definition of poverty."

o In the April 1966 Social Security Bulletin, Orshansky wrote, "...no upward adjustment was made [over the 1959-1964 period] in either of the [poverty] measures to take account of the higher standard of living that a rising real income makes possible for the majority....The difficulties [in setting the poverty line] are increased when the definition is to be used to measure progress over a span of time. Statistical nicety will be better served if the criterion selected remains invariant. The realities of everyday living suggest it cannot be--at least not for very long. Though the change in consumption patterns from any one year to the next might be minuscule, over the long run the upgrading that goes with the developing United States economy will be too great to be ignored. Research in consumer economics is not yet at a stage precise enough to specify just how long the long run is."⁽¹³⁶⁾ (The context indicates that "upward adjustment" refers to a raising of the real level of the poverty line--not just adjustment for price changes.)

o In a January 1967 draft, Ida Merriam wrote, "It is easy to observe that poverty in the U.S. today cannot meaningfully be defined in the same way as in the U.S. of 1900 or in India today. It is more difficult to project forward when and by what amounts the measure of poverty will need to be changed in the future. Yet obviously today's measure, even if corrected year by year for changes in the price level-- the purchasing power of money--should not be acceptable twenty, ten or perhaps even five years hence."⁽¹³⁷⁾

o In March 1967, responding to a comment on a paper of hers, Lenore Epstein said, "There would seem to be real merit in concurrent use of two measures [of poverty], one that changes with productivity and, for periods of five or at the very most ten years, one that changes only with price level....I would question using any budget-type measure-- whether constructed in great detail or more roughly as the SSA index--that is adjusted only for price change for a period as long as that from 1947 to 1965...."⁽¹³⁸⁾

o In November 1967, Orshansky wrote, "There must be a framework for adjusting a poverty line...for changes over time in the level of economic activity and the resultant rise in wages and general standard of living."⁽¹³⁹⁾

o In a December 1967 professional paper, Ida Merriam wrote, "An acceptable social minimum is obviously related to the general level of affluence of a society. In a dynamic economy it must therefore change over time. It is easy to reach agreement that what was an appropriate poverty measure in 1900 or 1933 is no longer relevant. It is also possible to get agreement that an acceptable social minimum in 1985 will be higher than today. It is difficult to find a satisfactory method of gradually moving the level up from its present to a hypothetical future position....Thus far the poverty index has been adjusted only for price changes. The need for a more substantial upward adjustment of the index level has been noted by many commentators. One solution that has much to recommend it would be two concurrent indexes. The SSA poverty and low-income indexes could be continued--adjusted only for changes in purchasing power--through say 1969. A second set of indexes could be adjusted to reflect productivity as well as price changes. The second set could start from the 1959 level...or the divergence could start in 1963 or later."⁽¹⁴⁰⁾

o In the March 1968 Social Security Bulletin, Orshansky wrote, "...as time goes on, a continuing rise in economic activity will make it difficult to avoid raising the poverty line."⁽¹⁴¹⁾

In addition to this concern, SSA personnel were also concerned about the fact that prices in general (as measured by the Consumer Price Index) had been rising more rapidly than the food prices (the per capita cost of the economy food plan) which were then being used to adjust the poverty thresholds for inflation each year; in other words, as measured by the CPI, the thresholds were actually decreasing in real terms.(142) In particular, the per capita cost of the economy food plan--and thus the poverty thresholds (at the economy level)--did not change at all from 1963 to 1964 and again from 1966 to 1967, even though overall consumer prices did rise during those periods.⁽¹⁴³⁾

Because of its concerns about the poverty-line/standard-of- living issue and the price index issue, SSA seems to have made a tentative decision early in 1968 to adjust the poverty thresholds to bring them more in line with the higher general standard of living by using data from the 1965 rather than the 1955 Household Food Consumption Survey. The specific step proposed to do this was to use the recently revised version of the economy food plan--updated on the basis of consumption data from the 1965 survey--to recalculate the (economy-level) thresholds; the revised economy food plan cost 8 percent more than the unrevised (1955-survey-based) plan, so the thresholds would have been raised in real terms by that percentage. (The revised low-cost food plan cost 4 percent more than the unrevised low-cost food plan, which would have resulted in a 4 percent rise in the near- poverty or low-income thresholds.) One argument advanced for using the revised food plans was that they rather than the unrevised food plans were being used in the family budgets which the Bureau of Labor Statistics had just begun to release.⁽¹⁴⁴⁾

On April 26, 1968, SSA convened an interagency meeting of technical staff from the federal agencies with an interest in poverty. (SSA personnel present included Merriam, Lenore [Epstein] Bixby, and Orshansky.) In a May 3 memo, Merriam described this as an "informal group called together to advise ORS-SSA on changes to be made" in the poverty and low-income thresholds for 1967. SSA presented to the group its proposal to use the revised food plans to recalculate the poverty and low-income thresholds, and the group agreed with the proposal. (In June 21 and July 23 memos, Merriam described the "decision" [presumably meaning the final decision] to revise the thresholds as having been made after consulting with the agency representatives at the meeting.)⁽¹⁴⁵⁾

Other subjects were also discussed at the April 26 meeting. One was an SSA proposal to adjust "the nonfood portion" of the poverty thresholds for inflation by "the total CPI minus medical care," while presumably continuing to use the per capita cost of the economy food plan to adjust the food portion of the thresholds. The interagency group favored a shift to CPI indexing of the thresholds, but decided that it should not be implemented for processing data for the current year (1967).⁽¹⁴⁶⁾

The specifics of the SSA proposal to adjust the thresholds for inflation are of interest because they imply that SSA viewed the poverty thresholds as not including any implicit allowance for medical expenditures--in other words, that SSA assumed that a family unit with a poverty-level income should not have to meet medical costs out of that amount of cash income. (Presumably SSA assumed that such a family unit would receive charity medical care, care under the Hill-Burton Uncompensated Services Program, or care under the relatively new programs of Medicaid or Medicare.) After the original version of the present paper was completed, Orshansky confirmed this inference (personal communication, April 15, 1993).

Also discussed at the April 26 meeting was an idea to replace the poverty threshold multiplier of 3 (derived from the 1955 Household Food Consumption Survey) with a higher multiplier derived from the 1965 survey⁽¹⁴⁷⁾, although this idea was not part of the formal SSA proposal for revising the thresholds. Together with the use of the revised food plan, this higher multiplier would have resulted in poverty thresholds 25 to 30 percent higher than the existing thresholds. "...it was agreed that it would be unwise to change this factor at this time....it was thought that additional data and analysis should underlie any new decision on this aspect of the index" (Merriam memo, p. 2).⁽¹⁴⁸⁾ "There was also some sentiment for moving over to a 'relative' rather than an 'absolute' definition [of poverty], or as a possible compromise, for raising the standard periodically (perhaps every 4 or 5 years) to reflect general increases in the standard of living enjoyed by the average U.S. family" (Stein memo, p. 5).⁽¹⁴⁹⁾

According to the Stein memo about the April 26 meeting (pp. 4-5), "Representatives of USDA [the Department of Agriculture] expressed dissatisfaction with the present ratio which indicates that farm families require only 70 percent of the money income required by nonfarm families to attain a presumably equivalent level of living. The studies by USDA imply that the ratio should be raised to 85 percent. This proposal was thought to need more study and was not suggested for adoption this year." According to the Merriam memo (p. 2), "The representatives of the Department of Agriculture confirmed the fact that the 1965 [Household Food Consumption] survey data provide no reason to change the 30 percent home produced food ratio used for farm families."⁽¹⁵⁰⁾

SSA started to implement its decision to revise the poverty thresholds on the basis of the revised food plans, and by about mid-June had poverty population figures for calendar years 1966 and 1967 on both the old and the revised basis. However, SSA's informal agreement with the technical representatives of other agencies began to come unraveled about mid-June. (The approximate timing can be inferred from a June 21 memo of Merriam's.) "The first use of the new poverty series was to have been in the White House Report on the Negro. The Census Bureau proposed to issue a press release giving the 1967 poverty estimates on the old and the new series basis a day or two in advance of the release of the White House report. In the process of review and clearance of this release, the whole question of whether there should be a new series was raised. (For 1967, the new series count of the poor is about 2.8 million higher than the old series count.) Several agencies, notably OEO and the Council of Economic Advisers, repudiated the position taken by their representatives on our advisory group, and argued against publication of the new series figures. The issue became intertwined with questions regarding the effect of a new imputation procedure used by Census to estimate 1967 income (conceptually unrelated, but affecting all data tied to income), and with questions relating to the preferred method of adjusting the poverty index for price change" (July 23 Merriam memo, p. 1). "A difficulty at the time was that the Office of Economic Opportunity had used the poverty index for operating purposes in a number of its programs and a change would have affected budgets and/or rules and regulations to a serious extent. This kind of difficulty will be faced by any Administration and any Government agency that uses this or any similar poverty index for operating purposes. There is no easy resolution" (February 5, 1969, attachment, p. 4).⁽¹⁵¹⁾

On July 16, 1968, the head of the Office of Statistical Standards in the Bureau of the Budget (BoB) sent a letter to the head of the Census Bureau directing that there be "no change in the criteria for computing the poverty 'thresholds' for income year 1967"--in other words, that the 1967 thresholds be based on the unrevised (1955-survey-derived) food plans. The letter spoke of "a number of unresolved conceptual issues as well as technical problems and limitations of data. Tabulations of the 1965 food budget information are not yet available for the full year. More important, a revision in the recommended nutritional standards, which are basic to the whole system, is likely in the near future." The letter also said that BoB would appoint a task force "immediately" to start "[i]ntensive work...as quickly as possible to develop concepts and technical information required to re-evaluate the poverty thresholds for future use."⁽¹⁵²⁾

Merriam noted that the July 16 BoB letter, overruling SSA's decision to revise the poverty thresholds and announcing the appointment of a task force to reevaluate the thresholds, raised the question of which federal agency should be assigned primary responsibility for the thresholds. "From 1963 until this year, the SSA has produced the poverty (and low-income) cut-off points. These have been published in the Social Security Bulletin and have been made available to others in more detail on request. The arrangements, however, have been informal--SSA wanted the data for its own research, bought tabulations from the Census [Bureau] year after year, developed special tabulations for the Council of Economic Advisers and others on request, and published analytic articles" (February 5, 1969, attachment, p. 5). "From time to time the Budget Bureau seems to expect SSA to continue in that role....The method of announcing the Task Force, however, would suggest a different interpretation" (July 23 Merriam memo, p. 2). Reviewing other agencies that might be considered candidates for being assigned responsibility for the thresholds, Merriam wrote, "OEO should not take over the function since it uses the poverty cut-off points for administrative purposes; the Census [Bureau] does not do the kind of analytic research that is required; BLS [the Bureau of Labor Statistics] has its own budget approach; the Council [of Economic Advisers] cannot carry on a continuing function of this kind" (July 23 Merriam memo, p. 2).⁽¹⁵³⁾ (The ultimate resolution of this question was that while the Census Bureau was given responsibility for publishing poverty statistics, no agency was given primary responsibility for maintaining the definition of poverty and doing research related to it.⁽¹⁵⁴⁾)

Members (including Ida Merriam) of an interagency Poverty Level Review Committee had been selected by late September 1968. A September 17 BoB letter (which the present writer has not found) spoke of setting up a technical support group for the Committee.⁽¹⁵⁵⁾ The Committee held its first meeting on October 2, 1968. All members of the Committee were federal agency employees except for Harold Watts of the University of Wisconsin's Institute for Research on Poverty, who had been hired by BoB's Office of Statistical Standards to serve as a consultant to the group. Merriam discussed how the poverty and low-income indexes