Gordon M. Fisher
(NOTE: The 78-page paper summarized here is a revision of a paper presented October 28, 1994, at the Sixteenth Annual Research Conference of the Association for Public Policy Analysis and Management in Chicago, Illinois. (For references, see the 78-page paper.) I am grateful to Robert Clark for helpful comments on an earlier version of this summary. [The full paper is available on the Census Bureau's Poverty Measurement Web site, at the specific address http://www.census.gov/hhes/povmeas/publications/povthres/fisher3.html. A two-page mini-summary of the U.S. evidence only is also available. The views expressed in this summary, the mini-summary, and the full paper are those of the author, and do not represent the position of the U.S. Department of Health and Human Services.])
One of the central questions in debates about poverty lines is whether they should be developed and updated in "absolute" or "relative" fashion. (Once established, an absolute poverty line is updated for price changes only, while a relative poverty line is updated for changes in the median or mean income or consumption of the general population.) In American debates on poverty lines, the tendency is to discuss theoretical arguments for using one or the other form of updating. This paper takes a different approach by focusing on the empirical historical behavior of poverty lines, particularly before the 1960's. (During that period, unofficial poverty lines were developed and then superseded after some period of time by new unofficial poverty lines.) This historical evidence, while not widely known today, can make a significant contribution to the debate.
This paper assembles an extensive body of evidence from the United States, Britain, Canada, and Australia showing that successive poverty lines developed as absolute poverty lines show a pattern of getting higher in real terms as the real income of the general population rises.(1) (A term used for this phenomenon in the U.S. is "the income elasticity of the poverty line.") In the U.S., this evidence includes:
Similar although somewhat less extensive evidence is available from Britain, Canada, and Australia.
From roughly 1905 to 1960, American budget experts developed a number of "standard budgets" at different levels of living. (A "standard budget" is a list of goods and services that a family of a specified size and composition would need to live at a designated level of well-being, together with the costs of those goods and services.) Standard budgets can represent standards of living higher than poverty or minimum subsistence. In 1966, Ornati analyzed the costs of about 60 standard budgets prepared during the 1905-1960 period, and classified them as being at "minimum subsistence," "minimum adequacy," and "minimum comfort" levels. Comparisons with other figures show that his "minimum subsistence" category corresponds to our concept of poverty.
A 1973 study by Kilpatrick showed that Ornati's minimum subsistence figures over this 55-year period rose 0.75 percent in real terms for each 1.0 percent increase in the real disposable income per capita of the general population. Another study of commodities rather than dollar figures showed that the Agriculture Department's economy food plan (the core of the Orshansky poverty line) included much more of almost every food than did the food components of pre-1929 minimum subsistence budgets. The number of rooms allowed for a family of a given size was higher in a 1960 minimum subsistence budget than in pre-1924 budgets, while the 1960 budget unlike a 1908 minimum subsistence budget included modest allowances for electricity and public transportation.
In 1993 I wrote a paper on U.S. poverty lines before 1965, examining some budgets covered by Ornati as well as other budgets and figures. The poverty/subsistence figures that I examined (like Ornati's budgets) were all derived as absolute poverty lines. Yet over time, these successive absolute poverty lines rose in real terms as the income of the general population rose.
In the U.S., the poverty/subsistence line as measured by the Douglas minimum subsistence level and the Orshansky poverty line was between 1.5 and 1.9 times as high in real terms for 1963 as it was for 1923. In Britain (as discussed in the section on that country), an unofficial poverty line for 1963 was 1.7 times as high in real terms as a poverty line for 1924. And in Canada, as measured by two different minimum budgets and Podoluk's original low income cut-offs, the low income/subsistence line was either 1.7 or 2.1 times as high for 1961 as it was for 1926.
Besides comparing the real values of earlier and later subsistence budgets, one can also demonstrate the income elasticity of the poverty line by using budgets to show that a given constant-dollar level which is considered to be at or below poverty during a later period (of higher general real income) was considered to be well above poverty during an earlier period (of lower general real income). I did that in my 1993 paper by finding five U.S. budgets or other standards during the 1901-1929 period which were at Ornati's minimum comfort level two levels above his minimum subsistence [=poverty] level and demonstrating that they were all approximately equal to Orshansky's poverty threshold in constant-dollar terms. One of these standards, from 1923 equal to 102 percent of Orshansky's threshold was described as being "the attainment of the highest class of wage-earners and the cynosure of the rest"; in 1965 and even more so in the 1990's, no reasonable person would use that phrase to describe a standard of living at that constant-dollar level.
Since 1946, the Gallup Poll has repeatedly asked the following question: "What is the smallest amount of money a family of four (husband, wife, and two children) needs each week to get along in this community?" The average response to this "get-along" question has been higher than the Orshansky poverty line, being quite close to Ornati's minimum adequacy level. However, it seems reasonable to assume that the relationship between the "get-along" amount and family income is a good indicator of how the public's perception of the poverty line would vary over time in relation to family income (if a "poverty" poll question had been asked). Half a dozen analyses have found that the "get-along" amount rises by between 0.6 and 1.0 percent for every 1.0 percent increase in the income of the general population. (The results vary, in part because the analyses used different measures of the income of the general population.)
Another significant (although neglected) source of evidence about the income elasticity of the poverty line is the common knowledge of experts on poverty lines and family budgets, as documented in quotations from their writings. These experts were quite familiar with the social processes that result in the income elasticity of the poverty line, and felt no qualms about recording their knowledge without needing to rely on formal quantitative studies.
For the U.S., the earliest of these quotations that I have found goes back a century and a half to 1841, when Unitarian preacher and social reformer Theodore Parker wrote, "What is a luxury in one generation, scarce attainable by the wealthy, becomes at last the possession of most men.... As society advances, the standard of poverty rises. A man in New England is called poor at this day, who would have been rich a hundred and fifty years ago." I found four more such quotations from the period between 1902 and 1919, seven between 1934 and 1959, and over a dozen from the 1960's.(2) One which illustrates with particular clarity the social processes underlying the income elasticity of the poverty line was written in 1938 by Daugherty: "A standard budget worked out in the [1890's], for example, would have no place for electric appliances, automobiles, spinach, radios, and many other things which found a place on the 1938 comfort model. The budget of 1950 will undoubtedly make the present one look as antiquated as the hobble skirt."
Some of the quotations make ironic reading in the light of subsequent history, as when Orshansky's boss Ida Merriam wrote in 1967 that "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....obviously today's [poverty] measure, even if corrected year by year for changes in the price level...should not be acceptable twenty, ten or perhaps even five years hence."
A number of poverty lines were developed on a budget basis in Britain before 1952. For later periods, British poverty studies generally follow the 1965 example of Abel-Smith and Townsend by using the National Assistance/Supplementary Benefits scale (benefit payment) rate as an unofficial poverty line. I have seen two British analyses showing that absolute poverty lines in Britain have risen in real terms during the twentieth century, and have found several additional items of evidence of this phenomenon. Recent conceptual work on British poverty lines by Veit-Wilson has shown that these poverty lines represent not one but two different standards of living, the "primary poverty" (P1) standard and the "Human Needs of Labour" (HNOL) standard, with the latter being higher; this means that some of the specific comparisons made in the two earlier analyses cannot be used to demonstrate the income elasticity of the poverty line.
One analysis showed that the unofficial P1 poverty line for 1963 was 1.75 times as high in real terms as a P1 poverty line for 1924, while the unofficial poverty line for 1973 was 2.35 times as high. The other analysis showed that the unofficial P1 poverty line for 1981 was between 2.3 and 4.1 times as high in real terms as a P1 poverty line for 1931.(3)
According to one analysis, Rowntree's HNOL poverty line for 1950 was 1.2 times as high in real terms as his HNOL poverty line for 1936. A budget-based HNOL poverty line for 1992 specifically designed to be a Rowntree-style poverty line was 1.8 times as high in real terms as Rowntree's HNOL poverty line for 1950.
One analysis showed that the unofficial P1 poverty line for 1963 was higher in real terms than Rowntree's HNOL poverty lines for 1936 and 1950. (This demonstrates the income elasticity of the poverty line by showing that a constant-pound-sterling amount that was well above the P1 poverty standard during an earlier period was equal to the P1 poverty standard during a later period.) The other analysis showed that the unofficial P1 poverty line for 1981 was between 1.5 and 2.3 times as high in real terms as Rowntree's HNOL poverty line for 1936.
A comparison with British national accounts data shows that the proportions by which the two poverty standards rose in real terms over various time periods were generally roughly equal to the proportions by which real gross domestic product per capita rose over the same time periods. The various comparisons P1-to-P1, HNOL-to-HNOL, and earlier-HNOL-to-later-P1 provide an extensive and overlapping network of evidence solidly demonstrating the income elasticity of the poverty line (or rather of both standards) in Britain over the 1924-1992 period.
The earliest known British quotation relevant to the income elasticity of the poverty line goes back to the late eighteenth century, when Adam Smith wrote, "By necessaries I understand, not only the commodities which are indispensably necessary for the support of life, but whatever the custom of the country renders it indecent for creditable people, even of the lowest order, to be without....necessaries [include]...not only those things which nature, but those things which the established rules of decency have rendered necessary to the lowest rank of people." Strictly speaking, of course, Smith was defining "necessaries," not "poverty." However, his concept of necessaries implies a definition of "poverty" that would be based not on an unchanging biological concept of subsistence but on whatever "the custom of the country" or "the established rules of decency" consider necessary. It is an irony of history that those today in Britain and America who aggressively identify themselves as disciples of Adam Smith are generally opposed to definitions of poverty that are consistent with their master's definition of "necessaries."
After Adam Smith, I found four British quotations relevant to the income elasticity of the poverty line during the 1876-1903 period, including the following frequently quoted 1890 statement by the eminent economist Alfred Marshall: "...every estimate of necessaries must be relative to a given place and time...." While this statement is not an explicit recognition of the income elasticity of the poverty line, it does clearly rule out absolute poverty lines, since absolute poverty lines are based on an assumption that estimates of "necessaries" do not vary over time.
I also found over ten British quotations between 1929 and 1951 and half a dozen quotations between 1954 and 1989 which showed knowledge of the income elasticity of the poverty line. Addressing the social processes underlying this phenomenon, Wedgwood wrote in 1929, "...in the search for material welfare, our modern civilisation under conditions of industrial progress is continually manufacturing new and previously unwanted sources of pleasure, so that the old luxuries become the new necessities, alike for those who can and those who cannot afford them." And in a 1936 statement with particular resonance for the American situation today, poverty researcher Bowley wrote, "...the opinion as to what constitutes poverty has changed in fifty years, and it is desirable to draw up a new definition for future use."
In the absence of published studies, I compared the constant-dollar values of four pre-1945 Canadian subsistence budgets or poverty lines with Jenny Podoluk's original Low Income Cut-Offs (LICOs) for 1961 (the measure subsequently adopted by Statistics Canada as an unofficial low income line). A poverty line for 1896 was roughly equal in real terms to 24 percent of Podoluk's LICO for 1961. Two different subsistence budgets for 1926 were equal to 47 and 59 percent of Podoluk's LICO for 1961, while a subsistence budget for 1939 was equal to 54 percent of that LICO. These figures show a general picture of poverty or subsistence standards rising in real terms over periods during which the Canadian population's real standard of living rose.
For the same pre-1945 period, I was also able to track down four above-poverty Canadian budgets or standards. All four of them were lower in constant-dollar terms than Podoluk's LICO for 1961. This demonstrates the income elasticity of the poverty line by providing examples of constant-dollar levels that were considered well above poverty/subsistence during an earlier period (of lower general real incomes) but were considered at or below poverty/low income during a later period.
After adopting Podoluk's LICOs as an unofficial low income line, Statistics Canada subsequently revised them several times based on data from successive Surveys of Family Expenditures, with most of the revisions resulting in higher new LICOs (in real terms). Because of these revisions, the LICOs themselves exhibit income elasticity over time. Revisions made in 1973 and 1982 that resulted in noticeably higher revised LICOs corresponded to even greater proportional increases in real average family income during the corresponding base periods. For the 1991 revision, a smaller real increase in the LICOs corresponded to a small increase in real average family income during the corresponding base period. A 1993 revision that resulted in very little real change in the LICOs corresponded to a small net increase in real average family income during the corresponding base period.
In a 1992 book, Sarlo, a Canadian conservative, put forward a deliberately low poverty line (developed as a standard budget) that was some 42 percent lower than the revision of Statistics Canada's LICOs then in use, as well as being 17 percent lower in real terms than Podoluk's original LICOs for 1961. Sarlo may be classified as an "anti-advocate" someone whose "goal [is] to push the level of the poverty line (or budget) below a currently accepted level." While his poverty line was lower in real terms than even the earliest LICOs, it was higher in real terms than pre-World-War-II Canadian poverty/subsistence standards for instance, 73 percent higher than one minimum budget for 1926. Over a long enough period of time, then, even the extremely low poverty line of an "anti-advocate" can serve as an example of the income elasticity of the poverty line.
At least since 1973, the Canadian Institute of Public Opinion has asked a question essentially identical to the U.S. Gallup Poll's "get-along" question. A 1989 article by Michalos presented responses to this question for the period 1973-1985. When figures in this article are deflated and analyzed, they suggest that the real "get-along" amount has some tendency to rise (or fall) when real median family income rises (or falls).
I found four Canadian quotations between 1965 and 1991 which showed knowledge of the income elasticity of the poverty line. One of them, in a description of the 1982 revision of the LICOs, commented about the advisability of adjusting a poverty line only for price changes: "...in the extreme, consider the number of families [today] below almost any unrevised low income cut-off established in the days of Charles Dickens...."
In the absence of published studies, I compared the real values of two 1960's unofficial Australian poverty lines with a budget developed "according to reasonable standards of comfort" in 1920 by the [Australian] Royal Commission on the Basic Wage. For two specific reasons, it seems safe to assume that the Commission's budget represented a standard of living noticeably above the poverty/subsistence level. The real value of this budget was equal to 82 percent of Australia's unofficial but widely used Henderson poverty line for 1966, and to 74 percent of a poverty budget developed by Halladay for 1969. These comparisons demonstrate the income elasticity of the poverty line by showing that a real monetary level that was considered to be well above poverty/subsistence during an earlier period (of lower general real incomes) was considered to be below poverty during a later period.
Since 1945, the [Australian] Morgan Gallup Poll has asked a question that is quite similar to the U.S. Gallup Poll's "get-along" question except that the descriptive phrase is "to keep in health and live decently" rather than "to get along." A 1989 paper by Saunders and Bradbury analyzed responses to this question for the period 1950-1988. They found that the average response rises in real terms by between 0.53 and 0.85 percent for every 1.0 percent increase in the real income of the general population, which they measured using three different variables.
The social processes underlying the income elasticity of the poverty line can be gleaned from quotations given in the paper summarized here. As technology progresses and the general standard of living rises, new consumption items are introduced. They may at first be purchased and used only by upper-income families; however, they gradually diffuse to middle- and lower-income levels. Things originally viewed as luxuries for instance, indoor plumbing, telephones, and automobiles come to be seen as conveniences and then as necessities. In addition, changes in the ways in which society is organized (sometimes in response to new "necessities") may make it more expensive for the poor to accomplish a given goal as when widespread car ownership and increasing suburbanization lead to a deterioration in public transportation, and the poor are forced to buy cars or hire taxis in order to get to places where public transit used to take them. Finally, the general upgrading of social standards can make things more expensive for the poor as when housing code requirements that all houses have indoor plumbing add to the cost of housing. In the light of these social processes, the only kind of American society in which it would be sociologically justified to have had the same fixed-constant-dollar poverty line since the mid-1960's would be a society in which there had been essentially no technological change or innovation since 1960.
Even though the income elasticity of the poverty line was well known among American poverty line experts before about 1970, the official U.S. poverty line was never raised in real terms to reflect rises in the general standard of living. The primary occasion when the official poverty line was not raised was in 1968-1969, when an interagency Poverty Level Review Committee was re-evaluating the poverty thresholds after the Social Security Administration had been forbidden to implement a decision to raise the thresholds by a modest 8 percent in real terms. The Committee's records show that the primary objection to a higher poverty line was the fact that it would have resulted in a higher number of people being counted as poor. (None of the programs then using poverty for eligibility were "entitlement" programs, so the effect on program eligibility was only a secondary consideration.) Having proclaimed a War on Poverty in 1964, the Johnson Administration was in 1968 able to boast of a three-year drop in the poverty population of 5.6 million persons. In that context it would have been politically embarrassing to have reported a 2.8 million "increase" in the poverty population resulting from raising the poverty line in real terms; too many people might have misinterpreted the "increase" as being the result of failed Administration anti-poverty policies, rather than as the statistical result of a redefinition of poverty.
Further insight can be gained by comparing these 1968-1969 events with events only half a decade earlier, when poverty lines were first adopted at the start of the War on Poverty. In January 1964, President Johnson's Council of Economic Advisers (CEA) adopted a quasi-official poverty line of $3,000 for families of all sizes. This poverty line was 19 percent higher in real terms than the low-income line of $2,000 (in 1948 dollars) for families of all sizes used in 1949 only fifteen years earlier by the Congressional Subcommittee on Low-Income Families (SLIF). Similarly, when Orshansky's poverty line was adopted as the federal government's new quasi-official poverty line in May 1965, it was 14 percent higher in real terms than a low-income line in a study by Lampman published in 1959, only six years earlier. Two contemporary analysts explicitly noted that the CEA poverty line was higher in real terms than the SLIF low-income line, and Lampman and others must have been aware that Orshansky's poverty line was higher in real terms than his 1959 low-income line. However, I found no record of any objection to the adoption of either the CEA or the Orshansky poverty line on the grounds that poverty is absolute, that a satisfactory estimate of minimum absolute needs had been developed fifteen or six years ago, and that it would be wrong and/or confusing to adopt a new poverty line that was higher in real terms than the earlier poverty line. The events of 1964-1965 thus show that American policymakers and analysts did not have a long-standing, principled objection to having poverty lines that reflected contemporary standards. The difference between the 1964-1965 events and the 1968-1969 events was due not to matters of principle but to matters of expediency. As Miller and Roby commented only a year after the decision not to raise the poverty line, "In the 1960's, unlike earlier periods, the budget-oriented estimates [of the poverty line] have not changed to keep up with changes in average styles of life. This break with previous practice is because of political, not conceptual, constraints."
Given that the income elasticity of the poverty line was well known among American poverty line experts before 1970, it is puzzling that many participants in today's poverty line debates are unaware of it. This change in knowledge seems to be connected with a change in the identity of the groups that develop and/or study poverty lines in the U.S. Before about 1965, the people who developed (and studied) poverty lines were largely advocates of the disadvantaged rather than theoretical social scientists; they included social workers, employees of state bureaus of labor statistics, labor union representatives, home economists, and employees of federal social agencies, with economists being only one of a number of elements in the mix. Relatively little of the pre-1965 poverty line literature that I have found was in traditional economic publications.
However, that situation changed with the beginning of the War on Poverty in 1964. Poverty studies became a distinct field as such, and economists began to get involved in poverty line studies in large numbers. People who had been involved in poverty line studies during the earlier period gradually retired and/or died during the 1970's and 1980's. As the earlier groups were gradually replaced by economists, it appears that the history and traditions of the earlier groups tended not to be transmitted to the newcomers; the newcomers did not perceive them as part of "our" history. As a result, much of the knowledge about the income elasticity of the poverty line was lost to those who are now studying poverty lines.
1. An earlier version of this material was given to the National Research Council's Panel on Poverty and Family Assistance; the Panel referred to this historical evidence on pp. 32, 33, 98-99, 103, 141, and 319 of its report (Constance F. Citro and Robert T. Michael (editors), Measuring Poverty: A New Approach, Washington, D.C., National Academy Press, 1995).
2. [Since completing the 78-page paper, I have found two more such quotations, one from 1914 and one from 1932.]
3. This analysis had a range of ratios rather than a single ratio because of uncertainties over the best price index to use for the 1938-1955 period, and because poverty lines for a five-person family varied depending on the ages of the children.
[Go to Further Resources on Poverty Measurement, Poverty Lines, and Their History
Last updated 12/16/05