Appendix A
Background and Specification of the OPM and the SPM
The intent of these two tables, prepared by panel staff, is to provide a reference document with specifications for the Official Poverty Measure (OPM) and the Supplemental Poverty Measure (SPM) through the changes made in 2021, including commentary and selected references.
- Table A-1 provides specifications for the OPM, with observations from the panel staff and document changes (virtually none) since the measure was officially adopted in 1969.
- Table A-2 provides specifications for the SPM, notes differences from the OPM and from the recommendations of the National Academies of Sciences, Engineering, and Medicine panel that produced the 1995 report Measuring Poverty: A New Approach, which laid the foundation for the SPM. The table documents the changes that were made since the first SPM poverty rates were released in 2011, which were implemented for estimates released beginning in 2021.
TABLE A-1 Official Poverty Measure (OPM) Specificationsa
Type of Measure/Updating Approach |
Economic: The lack of sufficient economic resources to obtain minimum levels of necessary economic goods and services, measured by comparing a monetary poverty threshold for a unit of measurement to the unit’s available resources over a period of time (typically a year, but OPM measures have been constructed for longer and shorter periods) Absolute: Updated for price changes only (although see Threshold Updating below) References: Fisher (1992); Fisher (1992); Interagency Technical Working Group (2021); National Academies of Sciences, Engineering, and Medicine (2019, especially App. D); NRC (1995) |
||
When Adopted | Developed by Mollie Orshansky for 1963; used by Office of Economic Opportunity; adopted by Bureau of the Budget in 1969 as official measure in Statistical Policy Directive 14 References: U.S. Census Bureau (annually): Consumer Income P-60 Publication Series and Poverty Thresholds |
||
Component | Current Definition | Commentary | Changes Since Inception |
Unit for which Poverty Measured | Family: Two or more people related by blood, marriage, or adoption living in same household Unrelated Individual: Person living alone or with other(s) not related to the individual in same household (includes cohabitors) Excluded: Residents of institutions (nursing homes, prisons, etc.); unrelated individuals under age 15 (including foster children) living in a household; homeless |
Does not recognize increase in cohabitation or sharing of household costs among unrelated people. | None |
Poverty Threshold Concept | Food Times 3: Three times the cost of minimum food diet in 1963, derived from ratio of all (after-tax) spending to food spending for families of 3+ persons in 1955 survey (a so-called “expert” budget method of establishing poverty thresholds) | Threshold concept is outdated: today, families spend 8 times the cost of food consumed at home or eaten outside the home | None |
Threshold Adjustments | Number: 48 thresholds varying by family size up to 9+ persons; number of children < age 18 up to 8+; and, for 1- and 2-person units, age of head < / > 65 Derivation: Thresholds for families of 3+ persons based on USDA Economy Food Plan costs for adults, children, and various size families (assuming economies of scale), with assumptions about ages of children in each family size, times 3; thresholds for 2-person families use a multiplier of 3.7; thresholds for 1-person units set at 80% of corresponding 2-person thresholds; thresholds for people age 65+ set at about 90% of thresholds for younger people |
Method to determine family size/composition threshold adjustments produced hard-to-justify variations; no adjustment for geographic variations in costs of living or for variations in housing benefits and costs associated with ownership status (owner/renter) | 1963: 124 thresholds, including farm-nonfarm status of family (farm thresholds set at 70% of nonfarm) and gender of family head (male head thresholds higher than female head thresholds) 1969: Farm thresholds raised to 85% of nonfarm thresholds 1981: Nonfarm thresholds used for all families; thresholds by gender of head averaged; thresholds for family size increased from 7+ to 9+ persons (and number of children from 6+ to 8+) |
Threshold Updating | CPI-U (flagship index published by the Bureau of Labor Statistics) | OPM is an absolute poverty measure conceptually, using an inflation index to adjust thresholds. Because the CPI-U (the inflation index used since 1969) overestimated inflation during the late 1970s–early 1980s, OPM thresholds increased ~13% from 1969 to 2019 in real terms vs. thresholds adjusted with a corrected CPI-U-RS | 1963: Updated by Economy Food Plan costs 1969: Updated by CPI-U |
Resource Measure (to compare to thresholds) |
Gross Before-Tax Regular Money Income Market Income: Wages and Salaries Returns from Assets (e.g., dividends, interest) Child Support and Alimony (for custodial parent) Private Disability and Retirement Cash Transfers: AFDC/TANF |
Gross money income concept relevant in the 1960s before expansion of government assistance through in-kind programs and tax credits but outdated today; captures effects of economic cycles but not of important programs or nondiscretionary expenses | None |
Treatment of Medical Care Costs/Benefits | Thresholds implicitly include small amount for medical out-of-pocket spending; resources ignore medical care benefits/costs | OPM ignores expansion of medical care benefits/costs since the 1960s | None |
Treatment of Taxes | Thresholds are after taxes (except for property and sales taxes implicitly included in the multiplier); resources are gross money income (inconsistent) | OPM ignores major changes in tax programs (e.g., EITC) that benefit low-income families | None |
Treatment of Work Expenses | Thresholds implicitly include amounts for work-related transportation/dependent care spread across all families, working or not; resources ignore them | OPM ignores increased need/spending by workers for dependent care, commuting | None |
Component | Current Definition | Commentary | Changes Since Inception |
Data Source | CPS-ASEC—sample of 100,000 households (includes noninstitutionalized group quarters and Armed Forces members living off base or with families on base; college students included in parental households; 100% oversample of Hispanic households beginning in 1976) interviewed in February–April following income reference year Known as the March Income Supplement prior to 2002 when the sample was expanded to improve state estimates of children’s health insurance coverage; in addition to increasing the March sample in states with high sampling errors for uninsured children, the expansion involved asking the supplement questions of one quarter of the February and April CPS samples (not including households also interviewed in March) and interviewing selected sample households from the preceding November CPS sample during the February–April period |
OPM measures have been constructed with many data sets—e.g., ACS for small-area estimates and SIPP for part-year estimates | See Data Source entry in Table A-2 |
Data Quality | CPS-ASEC obtains high unit response rates but much lower rates for income items (40% of income imputed); exhibits underreporting of many income types, even after imputation (which itself can create error) of amounts for respondents reporting receipt. Negative or very low incomes can also be misleading—e.g., for self-employed people who may have experienced a business loss but have an income flow from their businesses; college students supported by their families; and possibly others. Household and person response rates, although high, are declining, and coverage ratios compared with the Census are decreasing (post-stratification to population controls only partly corrects) |
Concerns about income data quality are major drawback of OPM (and any poverty measure based on CPS-ASEC); ACS and SIPP have similar quality problems References: Meyer et al. (2009, 2015); Meyer and Mittag (2018); Bee et al. (2015), Hokayem et al. (2015); Scherpf et al. (2015); Fox and Burns (2021a); Shantz and Fox (2018); Bollinger et al. (2019); Meyer and Mittag (2019); Mittag (2019); Rothbaum and Bee (2021); Wheaton et al. (2021) |
Percent income imputed and net underreporting have increased over time for many income sources, particularly transfers |
a Abbreviations used in table: ACS, American Community Survey; AFDC/TANF, Aid to Families with Dependent Children/Temporary Assistance for Needy Families; CPI-U, Consumer Price Index for All Urban Consumers; CPI-U-RS, CPI-U Research Series; CPS-ASEC, Current Population Survey Annual Social and Economic Supplement; EITC, Earned Income Tax Credit; SIPP, Survey of Income and Program Participation; SSI, Supplemental Security Income; USDA, U.S. Department of Agriculture.
TABLE A-2 Supplemental Poverty Measure (SPM) Specificationsa
Type of Measure/Updating Approach |
Economic: The lack of sufficient economic resources to obtain minimum levels of necessary economic goods and services, measured by comparing a monetary poverty threshold for a unit of measurement to the unit’s available resources over a period of time (typically a year) Quasi-relative: Updated for real changes in consumption of basic bundle of goods—originally FCSU estimated from the CE; currently FCSUti References: Meyer and Sullivan (2012); Wimer et al. (2016); Shaefer and Rivera (2018) |
||
When Adopted | Proposed by 2010 report of Interagency Technical Working Group, based on recommendations of 1995 National Academies’ report Measuring Poverty: A New Approach (MP); implemented by Census Bureau and BLS: first SPM measures published in 2011 (income year 2010) References: U.S. Census Bureau (annual), Supplemental Poverty Measure; Bureau of Labor Statistics website, Experimental Poverty Measures | ||
Original Differences from National Academies (1995) |
Thresholds/Threshold Adjustments: Originally, expenditures for all families with 2 children instead of 2 adults and 2 children (although latter still the reference unit); revised 3-parameter (instead of 2-parameter) equivalence scale for family characteristics/economies of scale; separate thresholds for renters with rent >$0, homeowners with a mortgage, and homeowners without a mortgage Updating: Originally 33rd percentile of FCSU instead of % of median; average of 5 instead of 3 years of CE quarterly data Reference: National Academies (2019) |
||
Component | Definition (2011–2020) | Commentary vis-à-vis OPM and Measuring Poverty (MP) | Changes Since 2011/Adopted in 2021 |
Unit for which Poverty Measured (Unit of Analysis) | Family (expanded definition): Two or more people related by blood, marriage, or adoption or by cohabitation and living in same household; includes foster children under age 22 and other unrelated children under age 15 living with a family Unrelated Individual: Person living alone or with other(s) not related to the individual in same household Excluded: Residents of institutions (nursing homes, prisons, etc.); homeless |
Re OPM: SPM recognizes increase in cohabitation; in 2010 the SPM estimated 124,199 units of analysis compared with 131,946 for the OPM (while the CE estimated 121,107 consumer units for 2010). Most of the decrease for the SPM compared to the OPM was due to treating cohabiting partners as parts of family units instead of as unrelated individuals. (At the same time, the SPM added about 1 million children under 15 to the poverty universe) References: Provencher (2011); Hokayem and Heggeness (2014) |
None |
Component | Definition (2011–2020) | Commentary vis-à-vis OPM and Measuring Poverty (MP) | Changes Since 2011/Adopted in 2021 |
Poverty Threshold Concept | Basic necessities: FCSU plus a little more; FCSU set at 30th–36th percentile of spending for families (consumer units, CUs) with 2 children from CE; “a little more” set at 20% of FCSU; 5 preceding years (20 quarters) of CE data used for estimation (each quarter updated to last quarter using CPI-U) (percentile range implemented to allow for 3 thresholds by housing tenure) CUs comprise: (1) all members of a household related by blood, marriage, adoption, or other legal arrangement; (2) financially independent persons, including people living alone, sharing a household with others, rooming in a private home or lodging house or in permanent quarters in a hotel or motel; or (3) 2+ people living together who use their income to make joint expenditure decisions. To be considered financially independent, the respondent must provide entirely or in part at least 2 of the 3 major expense categories of food, housing, and all other; 97% of CUs include the entire household; the CU definition for estimating thresholds is similar to the SPM definition for assessing poverty status |
Re OPM: SPM recognizes larger bundle of necessities; derives thresholds from actual spending instead of expert judgments Re MP: SPM uses 5 instead of 3 years of CE data to reduce sampling error and adjust thresholds with a lag; uses families with 2 children because they are a larger fraction of CE sample than families of 2 adults/2 children Re OPM: The initial 2010 SPM threshold (ignoring tenure status) for 2-adult/2-children families of $24,343 was 10% above the OPM threshold of $22,113, but this is a misleading comparison. Assuming the OPM threshold was $18,650 when converted to the SPM concept (which subtracts some budget items from resources), then the increase for the SPM threshold was 31%—near the top end of the 9–33% range suggested by Measuring Poverty (see pp. 53-56), but in the middle of the range of other budgets the 1995 panel examined. For 2019, the SPM threshold was only 24% greater than the converted OPM threshold References: Heisz (2019); Garner and Munoz (2021) |
2021 (implemented beginning with revised 2019 thresholds—see spm_threshold_200520.xlsx (live.com)):
|
Threshold Adjustments |
Family composition: 3-parameter equivalence scale used to adjust 2-adult/2-child threshold for differing numbers of adults and children, economies of scale:
Housing tenure: Separate thresholds developed for homeowners without a mortgage, homeowners with a mortgage, and renters paying rent, using average of 30th–36th percentile expenditures for each group Geographic price differences: ACS used to adjust thresholds for housing costs across geographic areas—specifically, 5-year ACS median gross rent for 2-bedroom units with complete kitchen and plumbing facilities estimated for 260 large MSAs, all nonmetropolitan areas within state (47), and combinations of smaller metropolitan areas within state (35); 342 adjustment factors applied to housing and utilities components of thresholds and results added to the other components |
Re OPM: SPM’s method for equivalizing family needs is an improvement on the original idiosyncratic method (although no equivalence scale can be said to be “truth”); recognizes differences in cost of housing across geographic areas (biggest item) Re MP: Subsequent research determined SPM 3-parameter scale gave more intuitive results than MP 2-parameter scale 5 (adults + 0.7 * kids)0.65-0.75; need became apparent for method to account for lower housing expenses of homeowners without a mortgage References: Family composition: Betson (1996); Social Metrics Commission (2020) Geographic price differences: Renwick (2011); Pacas and Rothwell (2020) |
2021: See Poverty Threshold Concept above |
Threshold Updating | 30th–36th percentiles of spending on FCSU reestimated each year from previous 5 years (20 quarters) of CE, updated to latest quarter with CPI-U | Re OPM: SPM is a quasi-relative measure on assumption that increases in FCSU spending lag all spending but exceed price inflation; intended to achieve gradual increase of thresholds in real terms in contrast to stasis of OPM thresholds Re MP: SPM felt to be more transparent by recalculating each year at 30th–36th percentiles instead of percent change in median FCSU |
2001:
|
Component | Definition (2011–2020) | Commentary vis-à-vis OPM and Measuring Poverty (MP) | Changes Since 2011/Adopted in 2021 |
Resource Measure (to compare to thresholds) |
After-Tax Disposable Income: Plus: Wages and Salaries Self-Employment Income (net) Farm Income (net) Returns from Assets (e.g., dividends, interest) Child Support and Alimony (for custodial parent) Private Disability and Retirement Cash Transfers: AFDC/TANF In-kind Transfers: SNAP, WIC, LIHEAP, NSLP, housing subsidies Minus: Net federal and state income taxes, FICA taxes |
Re OPM: SPM recognizes that tax and in-kind transfer programs are major sources of income support in the U.S.; recognizes that workers incur nondiscretionary costs; recognizes child support as nondiscretionary expense of the noncustodial parent; recognizes MOOP costs as nondiscretionary for the low-income population | 2021: Change from national average value for WIC benefits to state-varying values |
Treatment of Medical Care Costs/Benefits | Consistent Treatment: Thresholds do not include medical care expenditures Resources subtract health insurance premiums and other MOOP costs (copays, deductibles, other); MOOP costs estimated |
Re OPM: SPM captures effects of medical care policy changes that reduce (increase) MOOP spending, but does not explicitly value medical care or health insurance; medical care not included in thresholds because of difficulties of determining need for differently situated families References: Fox et al. (2015); Korenman and Remler (2016); Burkhauser et al. (2020); Korenman et al. (2019); Remler and Korenman (2021) |
None |
Treatment/Valuation of In-Kind Transfers | Partially Consistent Treatment: Thresholds include food expenditures with SNAP benefits but no other in-kind benefits Resources subtract SNAP, NSLP, WIC, LIHEAP, housing subsidies |
2021: Add imputed in-kind benefits to thresholds (consistent with resources): rental assistance, LIHEAP, NSLP, WIC (SNAP implicitly included in food expenditures) |
Treatment of Taxes | Consistent Treatment: Thresholds are after taxes (except for property and sales taxes included in FCSU spending and multiplier) Resources subtract federal and state income taxes and FICA taxes, using Census Bureau tax simulator |
None | |
Treatment of Work-Related Expenses | Consistent Treatment: Thresholds include nonwork transportation Resources subtract childcare expenses and other work-related expenses (e.g., work-related transportation, uniforms), capped by earnings of single parent or lower earner of two parents Other work-related expenses calculated as: weeks worked for each earner 18+ from CPS-ASEC * 85% median per earner weekly work-related expenses from most recent SIPP |
Re OPM: SPM recognizes increased need/spending by workers for childcare, commuting. For 2010 and 2019, not subtracting work expenses (childcare and other) from resources would have decreased the overall poverty rate by 1.5 and 1.5 percentage points, respectively (2.0 and 2.0 for kids <18, 1.5 and 1.6 for ages 18–64 and 0.3 and 0.45 for ages 65+) | None |
Data Source |
Thresholds: (CE), four quarters (annualized) for 5 years Resources: CPS-ASEC—sample of 100,000 households interviewed in February–April following income reference year (see Data Source Entry in Table A-1) |
Like OPM, SPM measures have been constructed with many data sets—e.g., ACS for small-area estimates and SIPP for part-year estimates; SPM measures have also been constructed by states and cities (e.g., California, NYC, Wisconsin) using ACS and state/city administrative records | Income questions: 1948: Wages and salaries; all other sources 1949: Self-employment asked separately 1950: Farm and nonfarm self-employment asked separately 1967: Social Security, interest, dividends, and rent asked separately 1968: Interest, dividends, rents, royalties combined; public assistance, unemployment/workers’ compensation asked separately 1975: SSI, AFDC, general assistance, private/government pension income asked separately 1980: 51 sources of income separately identified as to receipt including numerous noncash benefits (e.g., food stamps, housing assistance); up to 27 amounts could be recorded 2000: Whether paid for work-related childcare added 2010: MOOP costs added (premiums, other OOP) 2014: Retirement account withdrawals added; health insurance/MOOP spending questions improved (ACA Marketplace insurance added) |
Component | Definition (2011–2020) | Commentary vis-à-vis OPM and Measuring Poverty (MP) | Changes Since 2011/Adopted in 2021 |
Data Quality | CPS-ASEC obtains high unit response rates but lower rates for income items (40% of income imputed); exhibits underreporting of many income types, even after imputation (which itself can create error) of amounts for respondents reporting receipt; problem of misleadingly low or negative incomes; CE data also problematic | Like OPM, concerns about income data quality are major drawback of SPM (and any poverty measure based on CPS-ASEC); ACS and SIPP have similar quality problems | See Data Quality entry in Table A-1 |
a Abbreviations used in table: ACA, Affordable Care Act; ACS, American Community Survey; AFDC/TANF, Aid to Families with Dependent Children/Temporary Assistance for Needy Families; BLS, Bureau of Labor Statistics; CE, Consumer Expenditure Survey; CPI-U, Consumer Price Index for All Urban Consumers; CU, consumer unit; FCSU, food, clothing, shelter, utilities; FCSUti, food, clothing, shelter, utilities, telephone, internet; LIHEAP, Low Income Home Energy Assistance Program; MOOP, medical out-of-pocket; MSA, Metropolitan Statistical Area; NSLP, National School Lunch Program; OPM, Official Poverty Measure; SSDI, Social Security Disability Insurance; SSI, Supplemental Security Income; SNAP, Supplemental Nutrition Assistance Program; WIC, Special Supplemental Nutrition Program for Women, Infants, and Children.