(From Focus 17:1, Summer 1995) Measuring poverty: A new approach It is dangerous to let a key social indicator become so frozen in place that, when societal conditions change, it can no longer adequately reflect what it was designed to measure -Measuring Poverty, p. 43 Poverty statistics routinely published by the Bureau of the Census are used in a variety of ways. The media use such numbers to compare trends in poverty over time and differences in poverty among such groups as the elderly and children. Government agencies and assistance programs use the official poverty threshold, or a multiple of that threshold, both in targeting services and resources to disadvantaged persons and as a standard of program eligibility. Academic researchers use the concept as a measure of individual and family well-being. Increasingly, however, there are questions about the soundness of the concepts and methodology from which the official numbers are derived. Given the importance of the official poverty measure, questions about its validity and utility cannot be ignored. In response to a request of the U.S. Congress, the Committee on National Statistics of the National Research Council in 1992 established a study panel to address the concerns about the poverty measure and to consider the conceptual and methodological issues in establishing standards for welfare payments to needy families. Anticipating the difficulties that the panel would face, Robert Haveman, Professor of Economics at the University of Wisconsin, commented: The official measure has become woven into the structure of both policy discussions and fiscal actions. Movements in the poverty measure are part of any observer's scorecard regarding the performance of political administrations and the health of the economy. . . . The work of the panel will be controversial. . . . The official measure is an arbitrary construct based upon a large number of implicit and explicit social judgments, regarding both economic needs and well-being. What is an "acceptable social minimum level of living"? What is to be counted in determining how well off are people who live in widely varying family types, regions, and neighborhoods? (Focus 14, no. 3 [Winter 1992-1993]: 28). The report of the panel has now been published. On May 11, 1995, the Institute for Research on Poverty (IRP) and the Office of the Assistant Secretary for Planning and Evaluation (ASPE) in the U.S. Department of Health and Human Services held a seminar that reported upon and evaluated its findings and recommendations. The papers from that seminar are presented in this issue of Focus. This introduction briefly reviews some of the intellectual and evaluation issues the panel addressed and summarizes its recommendations. Poverty defined The panel defined poverty as economic deprivation, that is, a circumstance defined by a low level of material goods and services or a low level of resources to obtain those goods and services. One is said to be "in poverty" if one's resources are below a threshold level for needed economic consumption and "not in poverty" if those conditions are not met. This concept of poverty is not the same as "welfare" or "well-being." Welfare is a term for certain government assistance programs or the resources that are transferred by those programs; an example is Aid to Families with Dependent Children. Well-being is a much broader term capturing the overall condition of a person, encompassing both economic and noneconomic attributes. People who are economically deprived may also be deprived in other ways-for example, they may be socially isolated or subject to a high risk of crime or abuse-but the concepts are not the same. The panel encouraged the development of indicators of other kinds of deprivation in addition to the measure of economic poverty. Adequacy of the current poverty measure The panel assessed how well the official U.S. poverty measure is serving as a barometer and benchmark for policy, research, and general public understanding of economic deprivation. It concluded that the measure is outmoded and no longer accurately characterizes differences in poverty among population groups, across areas of the country, or over time. A few of the panel's concerns are summarized below. The panel noted that the current definition of family resources, namely, gross money income, assumes that some families have adequate resources for their basic needs when, in reality, tax payments and other nondiscretionary expenses put them below the poverty line. Conversely, the current definition fails to count all of the resources--such as in-kind benefits--that other families have available to meet their needs. More specifically: * Because of the increased labor force participation of mothers, there are more working families who must pay for child care in order to earn income. Yet the current poverty measure does not distinguish between the needs of families in which the parents work or do not work outside the home, nor generally between the needs of nonworkers and workers (who typically have commuting and other work-related expenses). * Because of differences in health status and insurance coverage, different population groups face significant differences in out-of-pocket medical care costs, but the current measure does not take into account this variation. * Because the current measure defines family resources as gross money income, it does not reflect the effects of government policy initiatives that have significantly altered families' disposable income and, hence, their poverty status. Examples are the increase in the social security payroll tax, which reduces disposable income for workers, and the growth in the Food Stamp program, which raises disposable income for beneficiaries. Moreover, the current policy measure cannot reflect the effects of future policy initiatives that may have consequences for disposable income, such as changes in the financing of health care, expansion of the Earned Income Tax Credit, and efforts to move welfare recipients into the workforce. The panel also concluded that there are significant problems with how the poverty threshold, the line separating the poor from the nonpoor, is set: * The thresholds are the same across the nation, although significant price variations across geographic areas exist for such needs as housing. * The family size adjustments in the thresholds are arbitrary in many respects, and changing demographic and family characteristics (such as the reduction in average family size) underscore the need to reassess the adjustments. * Changes in the standard of living call into question the merits of continuing to use the values of the original thresholds updated only for inflation. Historical evidence suggests that poverty thresholds-including those developed according to "expert" notions of minimum needs-follow trends in overall consumption levels. Because of rising living standards in the United States, most approaches for developing poverty thresholds would produce higher thresholds today than the current ones. This is true of the original approach, which based the thresholds on a multiplier for other expenditures applied to the cost of a minimum food budget, because food expenditures are now a much lower proportion of the total. The panel concluded that the current official poverty measure will become increasingly unable to inform the public or support research and policy making if it is not revised. It recommended a new measure that more accurately categorizes the extent of poverty across groups, areas, and time. It based its recommendations on scientific evidence to the extent possible, but necessarily used judgment as well. The determination of a particular type of poverty measure and, even more, the determination of a particular poverty line are ultimately subjective decisions. The panel made clear at each step in the report the character and status of the scientific evidence and the role of judgment (one panel member dissented from its recommendations). To help choose among alternatives, the panel developed a set of criteria for a poverty measure: it must be publicly acceptable, statistically defensible, and operationally feasible. Public acceptability A concept that varies greatly from a generally accepted intuitive notion of what constitutes poverty would probably fail to gain political acceptance. The basic premise of the measure should accord with common sense. Statistical defensibility The measure must be logically consistent. The current official poverty threshold is an after-tax concept, but the annual computation of the proportion and characteristics of people in poverty uses a before-tax family resource definition; this is inconsistent. Also, a poverty measure must allow for reasonable comparative analyses across time, across places, across types of families, and across population groups. Analysts and policy makers want to be able to say something about the incidence of poverty compared with ten years ago; about its incidence in the Northeast or the Southwest; about its prevalence among children, among minority groups, among female-headed families. Operational feasibility Data must be available with which to implement the measure. Income and expenditures are concepts that are generally understood; they can be measured, and reliable data can be collected. They should be at the core of the concept and the measure of economic poverty. In developing its recommendations, the panel addressed an interrelated set of problems: 1. What should be the nature of the poverty threshold? 2. How should family resources be defined? Types of poverty thresholds The current official poverty thresholds were originally developed on the basis of the cost of a minimum diet multiplied by three to allow for shelter, clothing, and other basic needs (see box, p. 2). They represent a type of expert standard, in that the minimum food budget for each family type was based on the work of expert nutritionists at the U.S. Department of Agriculture. Since their adoption for official use in the 1960s, the poverty thresholds have only been updated for price changes. In other words, they were turned into an absolute standard, even though they were originally set relative to their time and place. (The USDA minimum budget or Economy Food Plan took into account the food-buying patterns of low-income families at the time; also, the multiplier of three was based on the spending patterns of the average family.) Absolute poverty standards do not reflect real changes in income levels or other changes in society that may be important to understanding poverty (e.g., the increase in work-related child care costs for families with children). Explicitly relative thresholds, in contrast, are developed by reference to the actual expenditures (or income) of the population. A typical approach is to select a cutoff point in the distribution of total family expenditures or income adjusted for family composition-say, one-half the median-and designate that dollar amount as the poverty threshold for a reference family. Relative thresholds are vulnerable to two criticisms: first, that the choice of expenditure or income cutoff is arbitrary or judgmental; and second, that they do not provide a stable target against which to measure the effects of government programs. Relative thresholds change each year in response to increases or decreases in real consumption levels, instead of remaining fixed in real terms. However, absolute standards developed according to expert standards of need do not differ so much as may be supposed from explicitly relative standards. When originally developed, they are always conditioned by the time and place of their adoption and inevitably involve elements of judgment (e.g., the USDA could have provided for nutritional needs with a higher-cost or lower-cost Economy Food Plan than the one actually specified). Also, it is rare for expert standards to be maintained in absolute terms for very long periods of time. Typically, an old standard is replaced after a period by a new standard that is higher in real terms. The wide acceptance of the Orshansky SSA poverty measure may have come about because both public perceptions and relative measures of poverty were in substantial accord back in the early 1960s that an income of about $3,100 was the appropriate poverty line for a four-person family at that time (see Figure 1). The panel's recommended poverty concept continues to base the thresholds on basic needs, specifically, food, clothing, and shelter, with a small amount added to allow for other needed spending. The panel argues for using expenditure data directly to develop the thresholds, instead of relying on expert standards. It also argues for an updating procedure that is quasi-relative: the thresholds would be updated each year for changes in spending on food, clothing, and shelter, as distinct from updating them only for price changes or for changes in total consumption. Definitions of family resources Family resources can be defined on the basis of income or on the basis of actual consumption or expenditures. (Some also argue for a "crisis" measure of resources that includes the value of asset holdings.) There are arguments for both income- and expenditure-based definitions; in practice, the availability of high-quality data is often a prime determinant of which definition is used. A basic principle for a poverty measure is that the threshold concept and the family resource definition should be logically consistent: for example, if child care and other work-related expenses are treated as a deduction from income on the ground that money so spent is a cost of earning income and is not available for consumption, such expenses should not be part of the poverty-level budget. An alternative would be to develop different poverty thresholds for nonworking families, working families without children, and working families with children. The panel argued that the approach of deducting nondiscretionary expenses from income is more feasible and more accurately classifies families by their poverty status. Under the official U.S. poverty measure, a family's poverty status is determined by comparing its gross money income to the appropriate threshold. Whereas there could be advantages to adopting an expenditure-based measure instead, the U.S. government does not possess the data resources to do so. The sample size of the relevant official survey, the Consumer Expenditure Survey (CEX), is too small to provide reliable poverty measures for population groups or by geographic area. In contrast, the United States does have large, well- developed surveys for measuring income. Thus, the panel saw no alternative to the current income-based definition of family resources. Members did, however, believe that the concept of gross money income should be modified in order to be consistent with the proposed threshold concept. Family resources, they argued, should take into account the value of in-kind resources that are available for consumption, and should deduct required expenditures. In-kind resources include food stamps, subsidized housing, school lunches, and home energy assistance. Expenses to be deducted include out-of-pocket medical care expenditures and health insurance premiums, income taxes and social security payroll taxes, actual child care costs up to a cap for working parents, an allowance for other work-related expenses, and child support payments to another family. (Currently, the latter are counted twice: once as income to the recipient family and again as income to the paying family, which is another inconsistency in the official measure.) Recommendations for a new poverty measure Basic elements of the current and proposed measures of poverty are summarized in Table 1. Redefining the threshold The official U.S. poverty thresholds, concluded the panel, should comprise a budget for the three basic categories of food, clothing, and shelter (including utilities) plus a little more. Data from the CEX should be used to determine a threshold for a reference family of four (two adults and two children) as a percentage of median expenditures by such families on these items. This sum should then be increased by a small additional amount for other needs, such as household supplies, personal care, and transportation unrelated to work. However, the threshold should not account for such nondiscretionary expenditures as taxes, child care and other costs of working, and out-of-pocket medical expenditures, which should instead be treated as deductions from income. The panel chose not to recommend a specific threshold with which to initiate the new poverty measure, believing that to be a matter of judgment. They did, however, suggest that a reasonable range for the reference family of two adults and two children would fall between $13,700 and $15,900 (in 1992 dollars). The lower number equals the expenditures for food, clothing, and shelter ($11,950) by families at the 30th percentile of all two-adult/two-child families, with a multiplier of 1.15 for other needed expenditures; the higher number equals the expenditures for food, clothing, and shelter ($12,720) by families at the 35th percentile of all two-adult/two-child families, with a multiplier of 1.25 for other needed expenditures. Updating the threshold Because poverty measures tend to reflect their time and place, the panel proposed a regular, but conservative, updating procedure that would adjust the thresholds for changes in consumption that are relevant to a poverty budget, rather than for changes in total consumption. Each year, the threshold would be updated to reflect changes in spending on food, clothing, and shelter over the previous three years and then adjusted for different family types and geographic areas. This updating procedure would automatically, over time, reflect real changes in the consumption of basic goods and services without the need for a periodic, and inevitably disruptive, readjustment of the level. It represents a middle ground between the current approach of simply updating the thresholds for price changes, which ignores changes in living standards over time, and the approach of updating the thresholds for changes in total consumption. Adjusting the threshold The panel recommended that the two-adult/two-child family threshold be adjusted for other family types by use of an equivalence scale. The scale proposed by the panel treats children on average as consuming 70 percent as much as adults and applies a factor to the number of adult equivalents in the family to account for the economies of scale that are available to larger families (e.g., they can jointly use many durable goods and buy food and other items in bulk). The panel suggests that the scale economy factor (which is a power in the formula) should lie in the range of 0.65 to 0.75. The use of such a scale would be an improvement over the scale that is implicit in the current official thresholds, which exhibit many irregularities and anomalies (e.g., in some instances the thresholds for families with children are higher than the thresholds for the same-size families made up of adults). The panel further recommended that the thresholds be adjusted to account for differences in the cost of a standard, low-cost rental apartment across regions of the country and among different-sized cities. Available data are not adequate to develop cost-of-living adjustments for all goods and services; however, decennial Census data make it possible to adjust the housing component of the poverty threshold for geographic differences in the cost of a standard rental apartment (including utilities). Research indicates that housing costs exhibit the widest variation across areas, and the panel recommended that a housing cost index be developed for nine geographic regions and, within each region, for several size categories of metropolitan area. Defining family resources The panel recommended that the resources of a family or individual that are compared with the appropriate threshold to determine poverty status should be consistently defined to include money and near-money disposable income. The definition should include money income from all sources, as well as the value of such in-kind benefits as food stamps and public housing. It should exclude out-of-pocket medical care expenditures, including health insurance premiums; income and payroll taxes; child care and other work-related expenses; and child-support payments to another household. The child care deduction should be capped and should apply only to families in which there is no adult at home to provide the care; the deduction for other work expenses should be a flat amount per week worked. Researchers in general agree about the appropriateness of such adjustments to income as deducting taxes and work expenses, which are a cost of earning income and cannot be used for consumption, and about adding the value of in-kind benefits that support consumption. The treatment of medical care benefits and costs is a more difficult issue. Trying to account for private and public medical insurance benefits in the same way as in-kind benefits for such items as food and clothing would greatly complicate the poverty measure and cloud its interpretation. Health care needs vary greatly: some people have high medical costs, some very few. Also, medical care benefits do not substitute for money income in the same way as do, say, food stamps. (A person who has an expensive operation does not have many more resources freed up for basic consumption than does a person who makes a visit to the doctor.) Hence, the proposed poverty measure does not include an allowance for medical expenses, either those that might be covered by insurance or those paid for out of pocket; for consistency, therefore, the proposed family resource definition does not add the value of health insurance. It does, however, subtract out-of-pocket medical care expenses from income: many people must pay out of pocket to obtain that insurance or to receive care, and such expenses reduce disposable income. The proposed poverty measure will reflect changes in health care policy that affect disposable income (just as it will also reflect changes in child care subsidies that affect net out-of-pocket child care expenses by working parents). For example, if changes in health care financing reduce out-of- pocket medical expenditures and thereby free up resources for food, housing, and other consumption, the proposed measure will show a lower poverty rate; the current measure will not show this effect. Conversely, the proposed measure, but not the current measure, will reflect health care policy changes, such as premium increases, that reduce disposable income. The panel also suggested that a separate measure of medical care risk be developed, which would assess the adequacy of health insurance coverage for the population. (People lacking health insurance are not necessarily poor if they have adequate income for such needs as food and shelter and are healthy during the year; however, they are definitely at risk should they become ill or have an accident.) Adopting the proposed measure The panel recommended that the U.S. Office of Management and Budget adopt the revised poverty measure as the official measure for use by the federal government. It also urged that the Statistical Policy Office of OMB institute a regular review, on a ten-year cycle, of all aspects of the poverty measure, to identify improvements in concepts, methods, and data needed to keep the measure in step with societal conditions and to maintain its utility as a barometer and guide to policy. Effects of the proposed measure Poverty rates under both the current and the proposed measures were estimated with data from the March 1993 Current Population Survey (CPS), supplemented with data from the Survey of Income and Program Participation (SIPP) and other sources. In one set of comparisons, the overall poverty rate was kept the same for both measures-14.5 percent in 1992. (For the proposed measure, this was accomplished by determining a two-adult/two-child family threshold that, together with a scale economy factor of 0.75 and all of the panel's other recommendations, gave the official poverty rate for the total population.) The results show important distributional effects on the makeup of the poverty population under the proposed measure: most strikingly, they show higher poverty rates for families with one or more workers and for families that lack health insurance coverage and lower rates for families that receive public assistance. The results also show higher poverty rates in the Northeast and West and lower rates in the South and, to a lesser extent, in the Midwest (see Figure 2). In another set of comparisons, the panel used the midpoint of its suggested range for the two-adult/two-child family threshold of $14,800. With this threshold, a scale economy factor of 0.75, and the other features of the measure, the poverty rate increased from 14.5 percent to 18.1 percent; with a scale economy factor of 0.65, the poverty rate increased to 19.0 percent. The changes in the resource definition increased the poverty rate more than did the changes in the thresholds. Had SIPP data been used exclusively, the panel estimated that the rate would have shown a smaller increase, from 14.5 percent to 15 or 16 percent (depending on the scale economy factor), because SIPP obtains more complete income reporting for lower-income people than does the March CPS. Use of the poverty measure in government programs How would changing the measure of poverty affect eligibility for government programs? Of seventy federal and federal-state programs that provided cash or in-kind benefits to people on the basis of an explicit test of low income in 1992-1994, twenty-seven programs linked their need standard for eligibility to the U.S. Department of Health and Human Services poverty guidelines, which are derived from the official poverty thresholds (Table 2). Use of the proposed poverty measure for programs that already use the current measure would more effectively target benefits to needy families. The proposed measure has an internally consistent equivalence scale by which to adjust the poverty thresholds for different types of families, it reflects geographic differences in the cost of housing, and its definition of family resources as disposable money and near-money income is consistent with the basic needs concept underlying the thresholds. However, not all federal agencies would find the measure equally appropriate to their needs. For programs such as the Food Stamp program that require a very detailed determination of both gross and net income in order to determine financial eligibility and benefit amounts, implementing the proposed measure would not complicate administration-indeed, that definition is similar to the one already in use. In contrast, other programs have a simple application procedure in which a crude measure of gross money income determines eligibility. Many of these programs-Head Start, for instance-provide an all-or-nothing service. Others charge recipients for services on a sliding scale, depending on the broad income-to-poverty ratio category into which the family falls. In these cases, full implementation of the proposed family resource definition could impose a burden on applicants and program administrators alike. However, the panel commented that there are ways to simplify the proposed definition for programs in which a simple application process is valued. Another issue is that the thresholds developed under the panel's proposed procedure may increase faster than thresholds that are simply adjusted by the Consumer Price Index. Clearly, there are budgetary consequences that flow from an annually updated measure, particularly for entitlement programs that must provide benefits for all applicants who meet the eligibility criteria. Updating the poverty thresholds for real growth in annual consumption of basic needs makes a great deal of sense for a statistical measure, but the design of government assistance programs must take into account many factors, only one of which is a statistical standard of need. Funding constraints, competing uses for scarce tax dollars, and the desire to provide incentives to low-income families to reduce welfare dependency may well dictate program eligibility levels that are lower than the statistical poverty thresholds. Data needs for the proposed measure The survey that has supplied the United States with its income and poverty statistics is the March income supplement to the CPS. The March CPS has served the nation well, but it is inherently limited in the extent and quality of the data that it can provide. As a supplement to a continuing survey of the labor force, its major focus is unemployment, not poverty. The March CPS currently obtains information on a family's previous year's income from a large number of sources, and it also asks about receipt of benefits from the major in-kind programs. However, it does not ask about taxes, medical care costs, child support, work expenses, or assets. Nor does it provide information for constructing poverty measures for periods other than a calendar year. To remedy these deficiencies in the March CPS and to improve the quality of information on family income and program participation, the SIPP was begun in 1983; it has largely achieved the goal of providing a richer set of higher-quality data. Poverty estimates calculated from SIPP, with more complete income reporting for lower-income families than in the March CPS, are several percentage points below the March CPS rates and comparable to estimates developed from the CEX that use a consumption or expenditure concept of resources. The changes to the family resource definition proposed by the panel will increase the data needed for measuring poverty. SIPP, with its focus on income, is in a position to respond to these needs (it already collects most of the required income and expense items). The March CPS, geared as it is to the main labor force survey, cannot so respond. Hence the panel recommended that SIPP should replace the March CPS as the basis of the nation's official income and poverty statistics. It also suggested modifications and improvements, in addition to those already slated for 1996, that would make SIPP even more suited to poverty measurement. The panel's recommendations extended beyond the mere substitution of one set of statistics for another, to address changes and improvements in data sources and research on more consistent and comprehensive poverty estimates from household surveys and the decennial Census. Among their recommendations were: * Decisions about the SIPP design and questionnaire should take account of the data requirements for producing reliable time series of poverty statistics using the proposed definition of family resources as disposable income. Population undercoverage, particularly of low-income minority groups, remains an issue of great concern. * To facilitate the transition to SIPP, the Census Bureau should produce concurrent time series of poverty rates from both SIPP and the March CPS by using the proposed threshold concept and updating procedure and the proposed definition of family resources. * The Census Bureau should routinely issue public-use files from both SIPP and the March CPS that include the Bureau's best estimate of disposable income and its components (taxes, in-kind benefits, child care expenses, etc.), so that researchers can obtain poverty rates consistent with the new threshold concept from either survey. * The Bureau of Labor Statistics should undertake a comprehensive review of the CEX to assess the costs and benefits of changes to the survey design, questionnaire, sample size, and other features that could improve the quality and usefulness of the data. The review should consider ways to improve the CEX for the purpose of developing poverty thresholds, for making it possible at a future date to measure poverty on the basis of a consumption or expenditure concept of family resources, and for the measurement of consumption, income, and savings. Finally, the panel confronted a range of other measurement issues and came up with specific recommendations: * The official poverty measure should continue to be derived on an annual basis and be supplemented by measures for shorter and longer periods. * The official measure should continue to use families and unrelated individuals as units of analysis, but should for purposes of poverty measurement broaden the definition of "family" to include cohabiting couples. The extent of resource sharing among roommates and other household and family members should be examined, to determine if it might be necessary further to modify the definition of the unit of analysis. * In addition to information on the number and proportion of poor people, the official poverty series should provide statistics on the average income and distribution of income for the poor. The count and other statistics should also be published for poverty measures in which family resources are defined net of government taxes and transfers, to help assess the effects of government taxes and transfers on poverty. Consequences of the panel's recommendations David M. Betson Department of Economics, University of Notre Dame How will the panel's recommendations affect the number and composition of those who are to be counted as poor? To answer this question, as well as to demonstrate that its recommendations could indeed be implemented, the panel conducted their own empirical analysis. This article summarizes the results of this analysis, which are presented in greater depth in Chapter 5 of the report. Data employed in analysis Although the panel recommends that the Survey of Income and Program Participation (SIPP) be utilized for poverty measurement, in their calculations they employed an extract of the 1993 March supplement of the Current Population Survey (CPS) public-use file, which was provided by the Census Bureau. The panel chose not to implement this specific recommendation, primarily because they lacked the time necessary to secure a SIPP extract and then impute variables, such as the amount of taxes paid, which are already imputed in the public-use CPS files. To implement the panel's recommended changes, numerous variables which are not asked on the CPS survey must be imputed to the data base. Currently, the Census Bureau provides imputations to many of these needed variables in the public-use file. The variables imputed by the Census Bureau that the panel utilized are: state and federal income taxes paid; earned income tax credit (EITC) received; federal payroll tax (FICA) paid; and the market values of the amount of food stamps, housing subsidies, and school lunches received by the household. To implement other recommendations, the panel needed to impute to the extract the amount of child care expenses incurred by the household; other work-related expenses; and the household's medical out-of-pocket expenses. The child care and work-related expense imputations were based upon data from a 1990 SIPP topical module. The medical out-of-pocket expense imputations were based upon data from the 1987 National Medical Expenditure Survey that had been aged to 1992. The panel did not feel that it could adequately impute the amount of child support paid by the household on the basis of information contained in the CPS; hence the recommendation to subtract this amount from the household's resources is not included in this analysis. Composition of the poverty population In order to focus upon the consequences of the panel's recommendations for the composition of the poverty population, the threshold for a family of four was chosen so that when all of the other recommendations had been utilized, the number of individuals who were poor would not change from the number who are currently counted as poor. In 1992, 36.8 million individuals were counted as poor by the Census Bureau when the $14,228 poverty line for a family of four is utilized. To achieve an equal number of poor, we had to adopt a threshold of $13,175, more than $1,000 lower. Although the poverty populations contain the same number of individuals, they are not the same two populations. One out of five individuals counted as poor under the panel's recommendations would be a "new" face in poverty. These 7.4 million new faces would not be counted as poor using the current Census Bureau methodology, but would be classified as poor under the panel's. Conversely, there are 7.4 million individuals who are counted as poor by the Census Bureau but who would no longer be viewed as poor using the panel's recommendations. Comparison of the two populations shows that, under the panel's recommendations, the poverty population includes 3.6 million more individuals living in families where the primary earner worked at least 48 weeks in 1992. This increase in full-year work does not come about because working families worked more weeks, but because 3 million fewer individuals lived in households where no member is working. Examining the 7.4 million new faces in poverty, we see that distribution of weeks worked almost exactly mirrors the distribution of work in the entire population. The panel's recommendations leave the total number of children in poverty almost unchanged. However, the type of family where they live does change. If we adopted the panel's recommendations, we would see more poor children living in families where both parents are present. However, the poverty rate of individuals living in single-parent families would continue to be high, and hence this group would continue to be overrepresented in the poverty population. The panel's recommendation to count many of the in-kind benefits yields a poverty population which is less likely to receive these benefits as well as other means-tested cash benefits (AFDC and SSI). Finally, the number of individuals in the poverty population that do not have health care insurance would rise. The number of poor The panel recommended that changes be made to the definition of the household's available resources and the threshold to be used in poverty measurement. Each of these recommendations would be expected to have a differential impact on the number of individuals counted as poor. To compare the impact of these individual recommendations, we computed the net number of individuals who would be poor if each recommendation was implemented separately. For example, if the net amount of taxes paid (the sum of state and federal income taxes, plus payroll taxes, minus the amount of EITC received) was deducted from the household's Census money income, then 1.2 million more individuals would counted as poor. The inclusion of the value of in-kind benefits in household resources reduced the poverty count by 4.2 million individuals. The marginal effect of adjusting the household's available resources for work-related child care expenses increased the count by 0.7 million, which was smaller than the marginal impact due to other work-related expenses; this latter resulted in an increase in poverty of 2 million persons. However, the largest marginal impact was due to subtracting the household's medical out-of-pocket expenditures; the result was an increase of 5.3 million individuals in poverty. The panel recommended three significant changes to the threshold concept. The marginal effect of adopting a geographic interarea price index would have only a minor impact of the overall number of poor (an increase of 0.2 million persons), but would have a significant impact on the regional and metropolitan distribution of the poor population. Although the panel recommended the adoption of a specific formulation for the equivalence scales, it believed that, in our present state of knowledge, it could not provide specific values for the parameters of the scale formula. To reflect that uncertainty, it suggested a range for the elasticity of the scales with respect to the number of adult equivalent persons in the family. This range was from .65 to .75. The marginal impact of adopting the scale formula using the .65 elasticity was to reduce the poverty count by 0.1 million, whereas the marginal impact of using the .75 elasticity was a reduction of the count by 1.9 million. The choice of the threshold for the reference family of four cannot be fully informed by "science." Although the choice of a value for the threshold is arbitrary, the panel did attempt to base its suggested range upon the amount that households do spend on the bundle of necessities composed of food, shelter, and clothing. The midpoint of the panel's suggested range was $14,800 in 1992. The marginal impact of changing the threshold to this level increased the number of poor by 1.8 million individuals. Using the midpoint of the range, $14,800, the total impact of all the panel's recommendations concerning the definition of available resources and thresholds was a net increase of 9.1 million persons in poverty, if .75 was assumed as the scale elasticity, and 11.4 million persons, if .65 was assumed. Summing the marginal impacts of the recommendations described above, one finds that this sum is less than the actual net change in the number of poor. For example, utilizing .75 as the scale elasticity, the sum of the marginal net changes is an increase of slightly more than 3 million persons, yet the combined effect of all the recommendations was 9.1 million. This difference implies a significant interaction effect between recommendations, an interaction effect that can be seen at work if one examines those individuals who are the "new" faces in poverty. Over 50 percent of those individuals who, by the current measure, are not considered poor, are reclassified as poor under the new measure, not by any single recommendation, but by the effect of a combination of various recommendations. Counting the impact of programs aimed at reducing poverty A serious shortcoming of the current poverty measure is that it does not recognize many programs which have been designed to fight poverty. One good example of such a program is the EITC. This refundable tax credit, which is given in the form of cash, is not counted as a potential resource for the family to meet its needs. If the panel's recommendations were adopted, then its receipt would be counted. To demonstrate the potential magnitude of counting this program, the panel estimated that, if the expansion of the EITC which is to take effect in 1996 had been implemented in 1992, 2.4 million individuals would been lifted out of poverty. The number of poor estimated using the current methodology would have been unaffected by this change in policy. Conclusions The panel sought to restore internal consistency to the methods by which we count the poor and at the same time to update the methodology to better reflect the realities of today's society. One standard by which we judged any potential recommendation was whether it was doable. We believe that by providing this analysis of our recommendations, we have demonstrated that the panel's recommendations can be implemented and that they do make a difference. If our recommendations are adopted, significant changes in programs aimed at poverty reduction will no longer go unnoticed by our poverty statistics. If our recommendations are adopted, we will find that individuals living in poverty are more likely to be working, more likely to be living in families where both parents reside, and less likely to be receiving cash or in-kind benefits from the government.# [Note: Tables and figures referenced in this document are not available on-line at this time.] Measuring Poverty: A New Approach Contents of the Report 1 Introduction and Overview What Is Poverty?/ The Official U.S. Poverty Measure/ Alternative Poverty Measures and Criteria for a Mea- sure/ A New Approach to Poverty Measurement: Recommendations/ Use of the Poverty Measure in Gov- ernment Programs 2 Poverty Thresholds Threshold Concepts/ Recommendations/ Expert Budget/ Relative Thresholds/ Subjective Thresholds/ Conclusions/ Implementing the Proposed Approach 3 Adjusting Poverty Thresholds Adjustment by Family Type/ Adjustments by Geographic Area 4 Defining Resources Overview and Recommendations/ Alternatives for Defining Resources/ Proposed Resource Definition 5 Effects of the Proposed Poverty Measure Data and Procedures/ Results/ Data Sources 6 Other Issues in Measuring Poverty Time Period/ Unit of Analysis and Presentation/ Indexes of Poverty/ The Limited Scope of Measuring Economic Poverty 7 Use of the Poverty Measure in Government Assistance Programs Recommendation/ Government Assistance Programs/ Using the Proposed Poverty Measure 8 The Poverty Measure and AFDC Determining Program Benefit Levels/ Determining State AFDC Standards of Need Appendices Dissent, by John F. Cogan/ Data Sources for Measuring Poverty/ The Interdependence of Time and Money/ Assistance Programs for People with Low Incomes