d d I 390 November 1, 1984 THE PROBLEMS OF MEASURING POVERTY INTRODUCTION Federal, state, and local governments last -year spent $107.5 billion on the major means-tested transf er programs--usually called we1fare.l According to the Census Bureau, 35.3 million Americans were "in poverty" that year. If this cash had been given outright to those poor, it would have averaged $3,048 per person, or a grant of $12,193 to a family of fo ur--well above the 1983 poverty.thresho1d of $10,1

78. Not only did this not happen but 40 percent of households below the official poverty line received no welfare benefits at On the other hand, one out of every five persons in America--about 42 million-- received at least one means-tested welfare benefit in 19

83. This includes 42 percent of black households (which had a 32.4 percent poverty rate) and 55 percent of female-headed households (which had a poverty rate of 36 per~ent Congressional Res,earch Se ryice memorandum, prepared by Carmen D. Solomon Analyst in Social Legislation, to the Senate Subcommittee on 1ntergove.rn mental.Relations, September 4, 19

84. The total includes only the ma.jor welfare programs (Aid to Families with Dependent Children, S upplemental Security Income, General Assistance, Earned Income Tax Credit, Medicaid Food Stamps, Housing and selected others) and does not include administra tive costs except for Medicaid and Food Stamps 1983 data, as yet unpublished, were obtained from the Census Bureau..

Cf 1982 data i'n Bureau of the Census, Characteristics of Households and Per sons Receiving Selected Noncash Benefits: 1982 (Current Populat.ion.

Reports, Consumer Income, Series P-60, No. 143 Table 10.

Benefit. data from" Bureau d'f the Census, Economic Characteristics of House holds in th'e United States: Third Quarter 1983 (Current Population Reports, Household Econom.ic Studies, Series P-70, No. 1 poverty rate daka. f.rom Money. Income. and Poverty S t atus of. Families and Persons in the Unkked' State Current Population Reports, Consumer Income, Series P-60, NO 1.45 2 Are so many Americans really impoverished? Should so many i be receiving welfare payments? The answers to these questions poverty for th e rest of this century are wobbly, in large part because Washington policy makers are not sure how to define poverty duced in 1965 and most recently revised in 1980 poverty on the basis of income only and sets income thresholds below which families are con s idered poor will lia've an iittportant influence on federal policies toward I Yet the essential statistics I There is an llofficialll definition of poverty, first intro It defines I I But the official definition was not designed to ascertain need and'thus has proved inadequate for various reasons, not the least of which is the known underreporting of income in census surveys and the fact that assets and in-kind income are ignored.

Because. of the difficulties with the official.poverty definition the federa l government has to modify it considerably when it tries to determine just who should be eligible for welfare. Some minor modifications result in a second, administrative, definition of poverty-the Office of Management and Budget guidelines--similar to th e Census Bureau's thresholds with a few exceptions To complicate matters further, however, each welfare program has its own rules for determining poverty status and eligibility.

Some of these rules exclude groups which the Census Bureau calls poor, but most have expanded eligibility far beyond the official Census thresholds. Finally, each state sets its own "standards of need" for eligibility for state programs.

The poverty thresholds are presumably an absolute standard below which it is impossible to main tain a decent living standard and above which, presumably, one can cope. The data base for determining basic subsistence needs, however, did not exist when poverty was defined and does not exist now. As a result, the official definition rests on arbitrary assumptions which surely overstate the extent of poverty, at least in terms of defining those in need of government help.

If policy makers are serious about fighting poverty, they must examine the official definition of poverty and improve it as a tool fo r designing programs flawed measure of need and therefore an inadequate guide to welfare policy declared, the goal was to eradicate hardship In the early 1960s, legislators and officials technically defined poverty in absolute terms'but dealt with it in r e lative terms. Relative poverty obviously is impossible. to eradicate--someone is always on the bottom. A relative approach thus makes the war on poverty unwinnable In its current state, it is a When the Great Society's War on Poverty was A new,poverty mea s ure, on the other hand, would tell policy makers not only who has a low income but also who needs public assistance. To do so, the concept of poverty should be defined 3 more sharply determininq poverty, on measuring income only; assets also should be inc l uded It should not rely, as has been the practice in The federal government should define the extent of public obligation to the poor and then define a minimal basket of goods and services necessary to maintain health. This market basket should be priced by the states, perhaps even with distinctions for urban and rural residents. The needs of families alike in size but widely varying in geographical areas and life circum stances cannot be determined by a single federal income threshold.

The "official pover ty rate" should .delete cash welfare from income in determining poverty status. T.his would permit a more exact assessment of how many Americans are impoverished. how much means-tested cash aid and what means-tested in-kind benefits (at market value) are g iven to the poor by government should be calculated. effectiveness and efficiency of the public involvement in aiding the poor Finally Only then can policy makers evaluate the MEASURING POVERTY: THE CENSUS BUREAU DEFINITION Economist Rose Friedman two dec a des ago asked What is poverty? Is it an income so low that it does not purchase enough food to allay hunger, as it was before 1800 and as it is in seven-tenths of the world today Or, is poverty eating hamburger when many others are eating steak? Is povert y having a cloth coat when others have fur coats, or having one bathroom when some have two or even three Is poverty always related to economic means In the absence of any scientific standards, is there any substitute for the judgment of the person setting the ~tandard Policy makers never fully discussed or answered these ques tions, and the present U.S. definition of poverty is based primar ily on the judgment of one person poverty standard adopted by the federal government was Mollie Orshansky, an officia l at the Social Security Administration.

Between 1963 and 1965, Orshansky devised a set of poverty thres holds to measure changes in the numbers and demographic composition of the poor over time. These quasi-official tools for judging the progress of the W ar on Poverty became official in 1969, and the Census Bureau was entrusted by the Office of Management and Budget with updating and revising them as necessary.5 The analyst who created.the Rose Friedman, Poverty: Definition and Perspective (Washington, D. C American Enterprise Institute, 1965 pp. 13-14.

U.S. House of Representatives, Committee on Ways and Means, Background Material on Poverty (WMCP:98-15 October 17, 1983, pp. 1-3. 'I 4 The original definition was calculated according to the prices of the .i .t.en?s n the Department of Agriculture's 1961 basic food budget and then multiplied by three chosen because a 1955 food consumption survey showed that the average expenditure on food by all families in the sample (com bining all income levels) was one-th ird of their after-tax income.

Orshansky used the average for all families even though she noted explicitly that ''poorer families generally devoted more than one-third of income to food, and those better off used less of their income in this way.If6 The t hresholds originally were adjusted for family size and age, with separate thresholds for farm families and female-headed families. Both of the latter distinctions were dropped in 1980.

The thresholds have been indexed to the Consumer Price Index since 196 9, but the underlying assumptions remain the same This multiplier was LIMITATIONS OF THE OFFICIAL DEFINITION Patterns of Spending According to the Orshansky poverty standard, one in five Americans was poor in 19

63. This was plausible, even though it coul d be argued that most Americans in that bottom quintile had better living conditions and possibilities for changing their status tha.n the typical Mexican, or even the average Soviet citizen. It was plausible because it is human nature to compare oneself with others in one's own society and not with some living standard worlds away..

Frie dman thought not. She questioned the.validity of using a multiplier based on the average of all families, when low-income families typically spend a far greater proportion of their income on food and thus achieve a nutritionally adequate diet despite thei r income level But was it accurate, even on its own relative terms? Rose The multiplier of three, in other words, injected a relative concept into the formulation of the supposedly absolute poverty measure. It presupposed that the poor do or should spend p ropor tionally as much on clothing, shelter, travel, and presumably concert tickets as the middle class. Using this assumption certain low-income families cannot have enough income left over Quoted in Friedman, op. cit p. 33.

The thesis that there is virtu ally no true malnutrition in this country and that poor nutrition in the U.S. is not primarily a function of income is supported by a number of otherstudies, including a 1977 report by the Department of Health, Education and Welfare. See Victor R. Fuchs H o w We Live (Cambridge, Massachusetts: Harvard University Press, 1983 pp. 65-66. 5 to buy sufficient food. Friedman estimated that the use of this miiltiplier, rather than one reflecting tb.e actva! spending patterns of low-income families, overestimated po verty by 100 percent in 19

62. Friedman herself calculated a 10 percent poverty rate rather than Orshansky's 20 percent, using Orshansky's own base criterion.8 injected a bias into the whole data series, affecting perceptions of what poverty means as well as the extent of the problem Acceptance of the higher estimate has obviously Indexing the poverty threshold to the Consumer Price Index CPI) after 1969 injected a further upward bias. The CPI measures the increase in cost of a specific basket of goods ove r time.

Indexing to the CPI thus ignores any substitution of one good for another by households to keep food costs down. If butter becomes too expensive, for example, a family might switch to margarine.

It might buy less beef and more chicken. These actio ns and the significant impact they have on a family's disposable income are completely disregarded by the CPI. Until a 1983 revision of the housing component of the CPI, moreover, the cost of buying a new home was given five times as much weight as rental costs in the CPI. Rapid inflation of home prices in the 1970s meant the CPI greatly exaggerated the actual increase in the cost of living for all those whq either already owned a home or, like most of the poor, who rented.

The poverty line, therefore, is now higher than it would have been had the original method of revision (multiplying a food budget times three) remained in place. For example, the 1983 poverty threshold for a family of four was $10,1

78. If the annual cost of the appropriate Department o f Agriculture food plan for four in 1983 ($3,036) were multiplied by 3, the poverty threshold in 1983 would have been $9,108, and significantly fewer people would have been "in poverty. I Non-Cash Income and Assets There are other problems with the offici al poverty measure.

It is determined on the basis of current annual cash income only and, for the mostpart, income that is regularly received, such as wages and salaries, self-employment income, Social Security public assistance, interest, rent, royalties, . alimony, veterans payments, and a number of other categories. Capital gains and one-time lump sum payments, such as life insurance, are not counted. Other resources, which are not income but certainly affect living standards, are ignored. Thus, borrowed money is not counted, nor are assets--even liquid financial resources--and in-kind benefits such as food stamps and Medicaid benefits.

Thus, a retired elderly couple owning a house and car and stocks and bonds was llpoorll if their Social Security and interest Friedman, op. cit pp. 34-35. 6 income were below $6,023 in 19

83. A college student who has moved out of his parents' home, and has less than $5,180 in annual cash scholarship and earnings income also is "poor even if he has loans or can fall back o n his parents' income. In one sense, these people may indeed be poor; they have little short term discretionary spending ability. But their situation is vastly different from that of an uneducated single mother of three who rents an apartment, has no car, and has impecunious relatives. Society's obligation. to this single mother surely must differ from its obligation to the student or home-owning retired couple. A measure of poverty that does not take posses sions or alternative income sources into account may have statis tical validity for certain purposes, but it does not describe poverty as traditionally understood. This factor contributes to overstating poverty.

Income Fluctuations Another quirk making the official poverty measure unsuitable for assessi ng need is its reckoning of income on an annual basis ignoring short spells of poverty. Example: If a single person without financial resources were unemployed and ineligible for unemployment compensation for half of 1983, and then worked in an 864-a-mont h job the second half of the year, his total income for the year would be $5,184--$4 above the poverty cutoff. On the other hand, if he were earning $860 a month--$5,160 a year he would have been counted as poor, even though for the last half year, and pre s umably into the future, he was earning a salary rate equal to 200 percent of the poverty level. This limitation of the poverty measure means that, at any given time in the year the poverty count--in the sense of those experiencing hardship may either be h igher or lower than the annual count. The direc tion of bias depends in large part whether economic conditions are improving or worsening.

Geographic Variations That the poverty threshold is based on national data may make it useful for handy reference and statistical purposes, but it also makes it unsuitable, without adjustments, for administer ing welfare on the basis of need. Obviously, there are differences in the cost of living in rural and nonrural areas, as well as in different geographic regions. A s now constituted, therefore, the poverty thresholds measure only income levels nationwide rather than degrees of hardship.

Unrelated Households Multiperson households are counted as a single unit only if they are related. Thus an unmarried couple living t ogether and sharing household expenses may both appear poor because they are counted as unrelated individuals, even.though their combined income may put them above the poverty threshold for a family 7 Underreporting of Income n 1 L CLAAapS &the iiicst 2Li w s problein with usirig official poverty data to establish actual need is underreporting of income. The income data on which the poverty statistics are based come from the Census Bureau's Current Population Survey, in which under reporting of income by res p ondents is well known. The Bureau's estimates for 1982 indicate underreporting of total aggregate income of about 11 percent. Income from Aid to Families with Dependent Children, Supplemental Security Income, and unemployment compensation probably was und e rreported by about 23 per~ent The underreporting of declared income is dwarfed by the non-reporting of income from the multibillion dollar underground economy. In addition to the countless household workers and handymen who frequently work for cash to avo id taxes, there is a large illegal economy. It would be difficult.to incorporate such information into official statistics, but it should not be ignored.

Aspects of Census Bureau methodology, however, may tend also to underestimat e poverty. Example: unrelated individuals are only included in the count if they are 15 or older. Result A poor child not living with his family is simply not counted. In addition, the poverty threshold is now determined on the basis of before-tax income, whereas the original food-budget-times-three formula was based on after-tax income.1 In sum, however, it is almost certain that the official poverty count greatly exaggerates the number of needy Americans.

Implicitly confirming this is the fact that 40 percent of the tpoor manage to get along with no government welfare benefits at all. It is clear then that the official measure of poverty is seriously inadequate as a basis for welfare policy.

MEASURING POVERTY: THE ADMINISTRATI'VE DEFINITIONS The Office o f Management and Budget (OMB) has been issuing poverty income guidelines since 1967 for use by federal agencies.'l These guidelines are now updated by the Department of Health and Human Services. They simplify slightly the basic Census poverty thresholds. The result is that the OMB thresholds are somewhat higher than official poverty lines for families smaller than four persons, and lower for larger families. In addition, the OMB thresholds increase the Census poverty line by 25 percent for Alaska and 15 p ercent for Hawaii, but not for other states.

Federal and other welfare programs, however, vary the guidelines Bureau of the Census, Money Income and Poverty Status, op. cit., p. 37.

U.S. House of Representatives, Committee on Ways and Means, op. cit., pp 3, 5 lo l1 Ibid., pp. 6-9. in determining e1igibility.l2 the OMB standard as the income eligibility level. For example free school lunches are available to all children from families with incomes below 130 percent of the poverty guidelines.

Food.Stamp pro gram, meanwhile, uses a monthly version of the annual.poverty level and judges eligibility on a month-to-month basis. And rules for the Special Supplemental Food Program for Women, Infants, and Children (WIC) enable local agency standards to be as high as 185 percent of OM3 guidelines.

Dependent Children (AFDC) and Medicaid-do not use federal poverty guidelines at all but are based on state standards of need many states, such standards are below the official federal poverty guidelines. Thus those whom the Census Bureau counts as poor may not be eligible for federally funded but state-administered aid.

Most major welfare programs, AFDC included, also set asset limits some quite low, so still more officially poor'l people are in eligible for welfare were det ermined, there presumably was a rationale for each deci sion. In states with comparatively low living standards and small budgets, for example, the 'Ineed'l threshold was found much lower than in wealthier states. The federal food programs, on the other h a nd, set generous eligibility levels in part to compen sate for low state standards--the intent was to increase the incomes of low-welfare-state recipients to levels more closely approximating national poverty standards Some programs use a percentage of Th e The two costliest welfare programs--Aid to Families with In As eligibility standards for each program and each state But the result of this plethora of definitions and regula tions is a mess of Byzantine complexity IPovertyIl can be any one of a variety of income levels, depending on the program.

Inadequate definitions and measurement ensure that much of govern ment welfare,spending is simply not targeted on what generally would be considered the poverty population, while some truly poor individuals may be ineligible for aid.

ALTERNATIVE POVERTY ESTIMATES In addition to giving an inaccurate picture of who are the poor, the official Census poverty figures do not give a true picture of the impact of government spending on the poor, because they ignore the s ubstantial growth of in-kind benefits. In fact two-thirds of government welfare spending so far has not been counted as income in determining who is poor 12 See Congressional Research Service Report No. 84-99 EPW, Cash and Non Cash Benefits for Persons wi th Limited Income: Eligibility Rules Recipient and Expenditure Data, FY 1981-83, compiled by Vee Burke Specialist in Social Legislation, updated June 18, 19

84. This 195-page report provides data on more than 70 programs 9 I1 The Census Bureau is trying to remedy this. Work on alterna tive poverty measures was begun at the Census Bureau during the Carter Administration by Timothy Smeeding. He and other welfare experts had argued that traditional measures of income and poverty needed to be changed because b y the early 1970s the market value of in-kind benefits had begun to exceed cash transfers. Smeeding's work culminated in the Census Bureau's Technical Paper No. 50 in 1982, entitled IIAlternative Methods for Valuing Selected In-Kind Transfer Benefits and M e asuring Their Effect on Poverty This work was requested by Congress. A U.S. Senate statement of September 16, 1980, noted Official poverty statistics published by the Bureau of the Census currently ignore billions of dollars of Government in-kind benefits , such as food stamps housing subsidies and medical care. The Congressional Budget Office has shown that including in-kind benefits in the income statistics would cause the number of people in poverty to decline to about 9 million as compared to official s t atistics showing nearly 25 million people in poverty The Committee considers it essential that official poverty statistics reflect, at the earliest possible date, the effects of in-kind benefits. Without such information Congress and the Executive Branch cannot be certain that Government transfer programs are properly targeted 13 The Reagan Administration has been continuing the work on this matter begun under Carter.

Some critics of the alternative measures have questioned whether Ilcashing out" in-kind b enefits and counting them as income to the poor gives an accurate picture of income distribu tion and hence relative poverty, if the in-kind benefits of the non-poverty population, such as employer-paid health insurance or the mortgage interest deduction, are not likewise counted as part of the income of other Americans. This argument might have some merit if the purpose of policy were income redistribution if the goal of an accurate poverty threshold is to enable govern ment to provide the poor with a dec e nt subsistence, then the income of the nonpoverty population is irrelevant. Policy. makers should be focusing only on whether the poor are able to meet basic needs But l3 Department of State, Justice, and Commerce, The Judiciary and Related Aeencies ADDro Driation Bill. 19

81. U.S. Senate. 96th Coneress. 2nd Session. v Siptember i6, i980, pp. 33-34. 10 How Is In-Kind Income Counted?

There are chree basic ways of looking at the value of in-kind income 1) Assigning the benefit its market value, or what the recipient would pay for the goods or services in the private government's cost in providing the benefit. And it is presumably what h is money is buying. But since the recipient of the benefit might not choose to buy the good or service in the same amount if I market. This has the advantage of being closely,related to the the measure of most interest to the taxpayer, because it indicate s I I I I he had the cash to pay for it, it would be imprecise to view the benefit as equivalent to the cash, as far as the'beneficiary is concerned 2) Assigning to in-kind benefits what economists call recipient1I or "cash equivalentll value, meaning the c ash value that the recipient puts on the benefit. The trouble is that though this is appealing in theory, it is almost impossible to design without using simplified estimates that distort the out comes 3) Developing a so-called poverty budget share value. This sets an arbitrary cap on the value of particular goods and services.

The cap is determined by the proportion of income spent on the item by poor persons in 1960 to 19

61. Example: if a poor family suallylI spends 30 percent of its income on housing, no matter how much'housing aid the government now provides for a poor family, it will not be valued at more than 30 percent of the family's budget, because this method assumes that one item cannot be substituted for another. According to this definition, in-kind benefits do not make a family better off overall if they are given more of a particular good than they usually consume question that this prompts is--IlWhat is the obligation of govern ment to provide.the poor with more of a good or service than t h ey usually consume if this does not make them better off More fundamentally, the appropriateness of using 1960-61 spending patterns, and the assumption they do not vary over time or by circumstance, are questionable as bases for calculation In the past tw o years, the Census Bureau has been releasing alternative poverty estimates based on these three approaches.

Their estimates are still experimental and tentative The three methods have been applied to three different benefit groupings food and housing; foo d, housing, and medical care; and food housing, and medical care other than institutional expenditures benefits, llreducesll poverty the most. For 1983, the alternative methods yielded poverty rate estimates ranging from 10.2 percent to 14 percent. The l1 o fficialIf poverty rate, on the other hand was 15.2 percent The The market value approach, particularly including, medical 11 Questions Raised by the Alternative Measures While the alternative poverty measures are a useful theoreti cal tool, they provide l imited guidance to policy makers. The new poverty measures are basically varying methods of placing a cash value on in-kind income. They do not challenge the basic assumptions about poverty contained in the official Census Bureau thresholds.

Policy makers should consider whether in-kind aid should be perceived as reducing poverty or simply making the poor more comfortable. If poverty is viewed not only in money terms but as a condition of dependency, it would be illogical to say that in-kind benefits decre ase poverty. In that sense, of course cash welfare payments do not decrease poverty either.

This is a major issue that the technicians who defined poverty ignored. Technicians should not be setting policy. Yet in being allowed to make arbitrary decisions about poverty defini tions, the technicians have defined the scope of a major policy goal.

CONCLUSION The Problems with Present Measurement Methods The popular image of Ilpovertyll is of migrant workers, black sharecroppers, unemployed and unskilled urban slum dwellers, the handicapped, and the elderly. In short, those incapable of helping themselves and in need of public or private charity original commitment to the War on Poverty was couched in such terms The The fact is that probably less than 3 percent of the popula tion find themselves in what has been called a "permanent under class" or llculture of poverty.Il14 Yet the official poverty rate in 1983 was 15.2 percent. The degree of income mobility in the United States is quite high, and the turnover ra t e in the poverty population reflects that. Evidence suggests that only between one-half and two-thirds of the officially poor in a given year are still poor the next year.15 Poverty for most is a temporary l4 Data from the Panel Study of Income Dynamics, c onducted by the Survey Research Center of the University of Michigan, show that during the 1969-1978 period, only about 2.6 percent of the population was persis tently poor (defined as poor in 8 of the 10 years and that from 1974 1978, only 1.8 percent we r e poor in all five years. Greg J. Duncan Years of Poverty, Years of Plenty (Ann Arbor, Michigan: The University of Michigan, Institute for Social Research, 1984 pp. 41 44. Oscar Lewis, who developed the "culture of poverty" idea in the 1960s also made a " r ough guess" in 1968 that only about 2.5 percent of the population (20 percent of the poor) then belonged to such a "culture Cited in Duncan pp. 46-47 Ibid p. 40. l5 12 condition, and is often the result of involuntary unemployment or personal decisions re g ardi n.9 marrj age divorce, childbearing moving out of a parent's household, and job changes. In the case of unemployment, economic growth is obviously the best poverty program. The responsibility of government with regard to volun tary=-or at least consc ious--personal decisions, however, is less clear.

The poverty thresholds developed by Mollie Orshansky in the early 1960s led to a definition of poverty that has exaggerated the number of America's poor. Intending to establish a minimum living standard for the poor, it incorporated a number of arbitrary assumptions about the spending patterns and needs of the poor that caused an upward bias in the measure. Indexing to the Consumer Price Index, an inappropriate measure for determining cost of.living increas es for the poor, added a further upward bias.

The number of the poor is determined by recourse to Census survey data. Participants in the surveys are known to underreport their income, by as much as 23 percent for welfare programs, and many do not answer i ncome questions at all. Consequently the Census Bureau must estimate such undisclosed income. And the Bureau has no way of correcting its calculation of poverty for the known underreporting.16 Moreover, the Census Bureau bases its determination of poverty status entirely on annual income, but common sense would indicate that poverty--the opposite of wealth--is a function not only of income but also of real estate and other possessions savings, and access to alternative income sources, none of which is coun ted by the Census Bureau in determining poverty.

Whatever other purpose they may have, the poverty statistics are applied to help the government determine who is needy and evaluate the success of programs in eliminating poverty. In fact, since they are ina dequate to the task of establishing need they are not the standard by which most public welfare spending is actually disbursed. Yet the government constantly uses the official statistics to measure progress, flailing itself unneces sarily for the apparent lack of success Ibid., pp. 40-41, provides some independent information, in addition to the Bureau's own estimates of underreporting, that underreporting has a significant impact on the poverty rate. For example, the Panel Study of Income Dynamics, with p r obably more complete income reporting than Census Bureau surveys, yielded data which led the Michigan group to estimate a 6.8 percent poverty rate for the U.S. in 1978, according to the official Census definition of poverty compared to the Census Bureau's own estimate of 11.4 percent for 1978. 13 I The so-called alternative poverty measures count in-kind benefits as income and thus lower poverty rates and qive a .more accurate picture of need and deprivation. But the basic assump tions about what poverty i s are no different from the present official measure. Both methodologies view poverty as a function of income only. Both assume that anyone thus determined to be poor is in need of a handout. And both assume a handout lifts the person out of poverty.

Towar d a Better Measure of Poverty A more appropriate and useful poverty measure would establish need as well as income levels. To develop such a measure, a minimum market basket of goods and services needed for mental and physical health should be identified. This could include compre hensive health insurance. This market basket would be the basis for establishing new poverty thresholds for different sized families in all the states. This would give the U.S. nationally defined objectives and standards but allo w states to determine the income thresholds necessary to achieve such standards. These might be lower or higher than present thresholds however, is that they would express an easily comprehensible concept of poverty and define the extent of public obligati on.

The thresholds should be indexed to a poverty market basket, not a general price index but there would be a national definition of poverty. The number of the poor would be counted by state threshold levels public welfare programs would use the same def inition of need What is critical There would be no national poverty level All Measuring Income The poverty threshold computations would delete means-tested cash aid from the definition of income, so that it would be possible to determine the underlying po verty rate--the extent of poverty before means-tested public aid. A formula should be devised, moreover, to determine the assets held by the income-poor.

The definition of income itself should be extended to include income not regularly received.

Measuring Impact A separate measure should calculate the amount and distribu tion of both means-tested cash aid and the market value of means tested in-kind income to assess the impact of poverty programs on people in poverty. Market value would be the pr oper measure because it would be simply a measurement of how much aid is given to the poor, both to individual households and in the aggregate.

This statistic could be called a public welfare impact measure not a poverty measure, and it would enable Americ ans to judge policies' success in providing the poor with a defined standard of living. This would then provide the basis for discussions of whether the poor could be provided with the same quality of life or level of services for less money, and how. 14 N ew measures of poverty would prov-de society with two critical categories of information: the extent of poverty; and the extent of the public's financial commitment to helping the poor. If poverty is viewed relatively, then eradicating it is an impossibil i ty. But if it were viewed absolutely, these new measures would gauge the problem, its extent, the assistance being given, and the effectiveness of that assistance. And policy makers could direct anti-poverty programs to those who need them S: Anna Kondrat as Schultz Fellow