About the Index

How the Index Calculates the Costs of Basic Needs

For each basic need, the Index calculates the costs for each household. Due to data limitations and conceptual reasons, the way these costs are calculated varies by basic need.

Housing

Data for housing costs come directly from the ACS. The ACS asks respondents detailed questions about their housing expenditures, including their rent, mortgage payments, and utilities paid. Many other researchers, including those at the Harvard Joint Center for Housing Studies (JCHS) and the National Low Income Housing Coalition (NLIHC), have used ACS data to analyze housing affordability.

We define housing costs as the following:

• For home owners (both owners with a mortgage and owners free & clear): We used selected monthly owner costs which is the sum of payments for mortgages, deeds of trust, contracts to purchase, or similar debts on the property (including payments for the first mortgage, second mortgages, home equity loans, and other junior mortgages); real estate taxes; fire, hazard, and flood insurance on the property; utilities (electricity, gas, water and sewer); and fuels (oil, coal, kerosene, wood, etc.). It also includes, where appropriate, the monthly condominium fee for condominiums and mobile home costs (personal property taxes, site rent, registration fees, and license fees).

• For renters: We used gross rent which is the contract rent plus the estimated average monthly cost of utilities (electricity, gas, water, and sewer) and fuels (oil, coal, kerosene, wood, etc.) if these are paid by the renter (or by someone else on the renter’s behalf). Gross rent is intended to eliminate differences arising from varying practices in including of utilities and fuels in the rental payment.

• For those with the tenure status “Occupied without payment of rent”: Given that these households have missing values for gross rent, we added their utility costs when available. This included their electricity, water, fuel, and gas.

Food

We calculate food costs using the “Low-Cost Plan” estimated by the U.S. Department of Agriculture (USDA) and county-level data from Feeding America’s Map the Meal Gap dataset.

USDA produces four food plans at successively higher cost levels. Similar to the family budget analyses from the Urban Institute, MIT, Economic Policy Institute, University of Washington, and Ludwig Institute for Shared Economic Prosperity, we use the USDA’s Low-Cost Food Plan, which is the second-lowest of the four food plans the USDA produces. The Low-Cost Food Plan represents the cost of food in the 25th to 50th percentile of food spending. The USDA plans are based on the assumption that families meet their nutritional requirements solely through home-cooked meals. Given that the average U.S. family spends about half of their food budget on food prepared outside the home, the USDA low-cost food plan likely underestimates the total food costs that many families, even of modest means, incur.

For our calculations, we start with the June 2023 USDA Low-Cost Food Plan data for the contiguous 48 states and Washington, D.C. We calculate the Low-Cost Food Plan for each household in the ACS based on the age and sex of household members, using the appropriate Low-Cost Food Plan figure provided by the USDA by sex and age group.

We follow the USDA’s recommendations to adjust for different household sizes:

• To calculate total food needs for a household, (1) sum the food needs for each individual in the household and then (2) apply the appropriate adjustment factor based on household size.

• For individuals in other size households, the following adjustments are suggested:
  • 1-person: add 20 percent
  • 2-person: add 10 percent
  • 3-person: add 5 percent
  • 4-person: no adjustment
  • 5- or 6 person: subtract 5 percent
  • 7- (or more) person: subtract 10 percent

Similar to the Urban Institute, MIT, Economic Policy Institute, United Way, University of Washington, and Ludwig Institute for Shared Economic Prosperity, we adjust the USDA’s Low-Cost Food Plan data for local variation in food prices by using Feeding America’s Map the Meal Gap dataset.

We calculated county-level food cost multipliers by:

• Estimating a national average cost-per-meal by calculating the average of all county-level cost-per-meal values from Feeding America’s Map the Meal Gap dataset.
• Calculating the ratio of each [county’s cost-per-meal] / [national average cost per meal], to create a county-level index.

The smallest geography available in the ACS public use file is a Public Use Microdata Area (PUMA). PUMAs are contiguous geographic areas with a population of at least 100,000 people. A PUMA may contain the entirety of a single county, a portion of a single county, or all or parts of multiple counties. We use the Missouri Census Data Center’s GeoCorr 2022 application to convert the county-level data into PUMA-level data that we can merge with the ACS data. We then multiply the Low-Cost Food Plan estimates for each household in the ACS by these PUMA-level multipliers to account for geographic variation in food costs.

Health Care

In calculating health care costs, we account for people’s health insurance premiums and other out-of-pocket medical expenses. The ACS does not ask about health care costs, but in its efforts to produce Supplemental Poverty Measure (SPM) estimates using ACS data, the Census Bureau created ACS SPM research files that include imputed variables from the Current Population Survey Annual Social and Economic Supplement (CPS ASEC) for a person’s health insurance premiums, a person’s medical out-of-pocket expenses, other than premiums, and a person’s Medicare Part B premium.

To calculate health care costs, we add up these three variables for each individual, then aggregate them at the household level to obtain a total estimate of how much a household has spent to meet its health care needs. We then divide the total health care costs figure by that household’s total post-taxes, post-transfers income.

Child Care

The index assumes that families require child care for children aged 12 or younger. Our main data source for child care prices is the U.S. Department of Labor’s National Database of Childcare Prices (NDCP). We used the NDCP’s county-level data on median annual prices for one child at market rate for center-based care in 2022 (the latest available). We use the Missouri Census Data Center’s GeoCorr 2022 application to convert county-level data into PUMA-level data, which we then merge with the ACS data. NDCP prices for infants, toddlers, and preschool children reflect the cost of full-time care. NDCP prices for school-age children (ages 5 to 12) reflect the cost of school-year (part-day) arrangements, such as before- or after-school programs. When NDCP data were unavailable, we used data from Child Care Aware of America’s 2023 report. In the ACS data, we assigned NDCP infant care prices to infants and 1-year-olds, toddler care costs to 2-year-olds, and preschool care costs to 3- and 4-year-olds. For children ages 5-12, we assigned them the cost of care for school-age children. We used the Consumer Price Index Research Retroactive Series (R-CPI-U-RS) to convert the 2022 NDCP data into 2023 dollars to match the ACS data.

While the NDCP provides cost estimates for both center- and home-based care, we used center-based rates. Since center-based care is more widely used by families and research suggests it yields improved educational outcomes compared to informal, home-based alternatives, we use center-based rates. This rationale aligns with the methodology employed by the Urban Institute in their modeling of economic security. In 2023, 66 percent of children aged 5 and younger who received non-parental care were enrolled in center-based arrangements.

How the Index Identifies Cost-Burdened Households

For each basic need, the index classifies people into one of four categories: comfortably meeting needs, getting by, burdened, or severely burdened. These classifications are primarily determined by the percentage of household income required to cover specific necessities. An individual is considered overall “burdened” if their household falls into either the burdened or severely burdened category for a given dimension.

These classifications are grounded in housing policy.. The U.S. Department of Housing and Urban Development (HUD) defines a household as cost burdened if housing costs (including utilities) exceed 30 percent of income and severely cost burdened if these costs exceed 50 percent of income. Our index uses these established parameters to define the burdened populations, while introducing a 20 percent threshold to distinguish between the remaining households: those “getting by” and those “comfortably meeting needs.”

To establish a consistent measure of cost burden across other basic needs, we apply the mathematical ratios inherent in the HUD housing framework to the food and health care dimensions. For both of these dimensions, the severely burdened threshold is set at 5/3 of the baseline burdened threshold (mirroring the 50-to-30 ratio), and the comfortably meeting needs threshold is set at 2/3 of the baseline burdened threshold (mirroring the 20-to-30 ratio).

U.S. government agency standards, established research, and consumer expenditure data inform the threshold baselines. While this uniform proportional approach forms the index’s basic framework, structural market realities necessitate methodological exceptions for certain dimensions. Child care, for instance, requires a distinct approach because realized household expenditures often understate the true cost burden. Families priced out of center-based care frequently rely on informal arrangements or reduce their work hours. Consequently, the child care thresholds evaluate the prevailing market cost of center-based care rather than actual out-of-pocket spending.

The construction of a multidimensional affordability index inherently requires methodological discretion. We are fully transparent regarding the threshold baselines selected and the rationales underpinning them. As we continue to refine this index in future iterations, we may update these parameters based on emerging literature, improved data availability, and ongoing feedback.

Housing

We define the housing cost burden thresholds as a percentage of household income as follows:

• Comfortably Meeting Needs: Housing expenses are 20 percent or less of household income.
• Getting by: Housing expenses exceed 20 percent but are less than or equal to 30 percent of household income.
• Burdened: Housing expenses exceed 30 percent but are less than or equal to 50 percent of household income.
• Severely burdened: Housing expenses exceed 50 percent of household income.

As we explained above, our index uses HUD parameters to define burdened populations, while applying a 20 percent threshold to distinguish between those “getting by” and those “comfortably meeting needs.”

Food

We define food cost burden thresholds as a percentage of household income as follows:

• Comfortably Meeting Needs: Food needs are 12 percent or less of household income.
• Getting by: Food needs exceed 12 percent but are less than or equal to 18 percent of household income.
• Burdened: Food needs exceed 18 percent but are less than or equal to 30 percent of household income.
• Severely burdened: Food needs exceed 30 percent of household income.

The 30 percent threshold for the Severely Burdened category is based on the administrative formula used by the Food and Nutrition Service (FNS) to calculate Supplemental Nutrition Assistance Program (SNAP) benefits. This formula expects families to contribute 30 percent of their net income toward food, with the program covering the shortfall between that contribution and a baseline minimum cost food budget.

While SNAP utilizes the Thrifty Food Plan (TFP) to calculate this baseline, our index utilizes the USDA’s Low-Cost Food Plan. The TFP assumes a strictly at-home food preparation model, which fails to reflect the time constraints faced by many working families. Furthermore, recent empirical analyses demonstrate that the TFP underestimates actual food costs. Using 2024 data, the Urban Institute found that the maximum SNAP benefit calculated using the TFP failed to cover the cost of a modestly priced meal in 99 percent of U.S. counties. Nationally, a modestly priced meal cost $3.41, 20 percent higher than the maximum TFP-based SNAP benefit. This is why we chose to use the Low-Cost Food Plan. Based on the June 2023 USDA benchmark for a standard family of four (two adults ages 20–50, two children ages 6–8 and 9–11), the Low-Cost Food Plan requires $1,073 per month, compared to just $973 under the TFP.

With the severe burden ceiling established at 30 percent, we scale the remaining threshold categories using the uniform proportional ratios applied to the housing and health dimensions of the index. The Severely Burdened threshold is calculated as 5/3 of the baseline Burdened threshold, while the Comfortably Meeting Needs threshold is set at 2/3 of the baseline burden.

To validate these derived thresholds, we compared the index parameters against an analysis done by the USDA Economic Research Service measuring the share of pre-tax income that households actually spend on food. The USDA found that in 2024, households in the lowest income quintile spent an average of 33.0 percent of their income on food, aligning with our Severely Burdened classification. Conversely, the middle-income quintile averaged 12.2 percent (falling almost between the Getting By and the Comfortably Meeting Needs categories), and the highest income quintile averaged 6.4 percent (well within the Comfortably Meeting Needs tier).

Health Care

We define health care costs thresholds as a percentage of household income as follows:

• Comfortably Meeting Needs: Health care costs are 7 percent or less of household income.
• Getting by: Health care costs exceed 7 percent but are less than or equal to 10 percent of household income.
• Burdened: Health care costs exceed 10 percent but are less than or equal to 17 percent of household income.
• Severely burdened: Someone in the household is uninsured, and below 500 percent of the official poverty line, OR health care costs exceed 17 percent of household income.

We chose 10 percent as the threshold for identifying households burdened by health care costs because it is widely used by health researchers. See, for example, Curchin and Schmitt 2026, Jackson and Keisler-Starkey 2024, Hill et al. 2019, Xu et al. 2019, and Banthin, Cunningham and Bernard 2008. We set the thresholds for the other burden categories using the same approach we use for the food and housing dimensions of our index. The severely burdened threshold is 5/3 of the burdened threshold, and the living well threshold is two-thirds of the burdened threshold.

We define a household as severely burdened if someone in the household is uninsured, due to a lack of health coverage, posing a systemic risk to the household’s economic stability and well-being. It puts the household at risk of having to liquidate assets, drain savings, or take on high-interest debt in the event of sudden medical bills not covered by health insurance. If someone is uninsured, but is above 500 percent of the poverty line, we did not count their household as severely burdened because we considered this income level to indicate that the person could likely afford health insurance. In 2023, 500 percent of the official poverty threshold for one person was $77,400.

Child Care

To determine the cost burden thresholds for the index, we used the Department of Health and Human Services’ definition of affordable child care as costing no more than 7 percent of a family’s income. We doubled that threshold to 14 percent to define households that are severely burdened.

A defining feature of child care costs is that they can be partially reduced through unpaid caregiving labor, which often requires a parent to work fewer than full-time hours. Some family budget calculators, such as MIT’s Living Wage calculator, adjust child care costs based on whether a family has one or two adults working. Our index incorporates this dynamic while explicitly acknowledging the implicit costs borne by parents who reduce their work hours to lower child care expenses—namely, lost wages, foregone retirement contributions, and stalled career progression.

We do this by categorizing a household as comfortably meeting its child care needs only if the unadjusted cost of care, calculated without assuming a reduction of work hours, is affordable (below 7 percent of household income). If the cost of child care becomes affordable only when adjusted downward for a parent working less than full-time, we categorize that household as “getting by.” Ultimately, this framework accounts for the reality that a parent may prefer to work full-time but is constrained by the unaffordability of child care.

We define child care cost burden thresholds as a percentage of household income as follows:

• Comfortably Meeting Needs: The unadjusted cost of child care needs is less than 7 percent of household income.
• Getting by: The unadjusted cost of child care needs is 7 percent or more of household income, but the adjusted cost (reduced if a parent works less than full-time) is less than 7 percent.
• Burdened: The adjusted cost of child care needs is 7 percent or more, but less than or equal to 14 percent of household income.
• Severely burdened: The adjusted cost of child care needs exceeds 14 percent of household income.

We define full-time work as working 35+ hours per week for 50+ weeks a year, or 1,750 hours annually. If all parents living with the child are working full-time, we assume full-time child care is needed. If one parent is working less than full-time, we prorate child care needs based on the number of hours the parent with the fewest hours worked. For example, if the parent with the fewest hours worked 500 hours in a year, we would allocate 29 percent of the cost of full-time child care to that family (500/1,750 = 29 percent). This linear assumption is likely to underestimate the burden for part-time workers, given that hourly rates for part-time care are almost always higher than the prorated equivalent of full-time care.

To differentiate between the “getting by” and the “comfortably meeting need” categories, we calculate the child care burden for each household in two ways. First, we calculate their child care needs as a share of total household income based on the number and ages of children without adjusting for whether a parent is available to provide care. For example, if child care needs are $10,000 for a household with $100,000 in income, we calculate the child care cost burden to be 10 percent. Second, we calculate the child care cost burden while accounting for whether a parent is available to provide care. For example, if a parent in that household works half-time, we assume the household’s child care needs are $5,000 rather than $10,000, or 5 percent rather than 10 percent of household income. In this example, the household would be classified in the “getting by” category since its unadjusted child care burden figure is above 7 percent of the household income, but the figure adjusted for parental work hours is below 7 percent of the household income.

In our index, we also wanted to account for the fact that some parents who are not working might not be available to handle child care duties due to having a disability or being enrolled in school. If a parent reported one of the following three disabilities in the ACS data, we assumed that this parent was not available to care for a child under the age of 13: difficulty remembering, concentrating, or making decisions; difficulty bathing or dressing; or difficulty doing errands alone. In terms of identifying student parents, the ACS data does not allow us to distinguish between part-time and full-time students. If a parent reported being enrolled in school, we treated that as the equivalent of working 864 hours annually. We calculated 864 hours by assuming a part-time load of 8 credit hours for 36 weeks of the year, about 8 instructional hours, and 16 hours of independent study, reading, and assignments per week. Assigning all enrolled students an 864-hour (part-time) equivalent does underestimate the child care needs of full-time student parents.

We also wanted to account for the possibility that a household’s child care costs could be reduced due to the household’s access to Head Start, public pre-k, and child care subsidies. The ACS data does not identify who receives free or reduced-cost government-provided child care. However, we can use the ACS data to identify likely Head Start and public pre-k participants by their age (3- and 4-year-olds) and their public school enrollment. For those who meet these criteria, we assume that 60 percent of their center-based child care costs are covered. That’s equivalent to receiving about 1,000 free hours of child care coverage per year, or about 6 hours of instruction per day over a 180-day school year. One limitation of this analysis is that we did not identify likely Child Care and Development Fund (CCDF) subsidy recipients. Not accounting for those subsidies overestimates the child care burden of the households that receive them. This limitation would be particularly important to address if more child care subsidies are made available. As of 2022, about 16 percent of children eligible for subsidies under federal rules received assistance. We hope to address this in future work.

The Cumulative Impact of Cost Burdens

We designed these thresholds to identify where the cost of a basic need begins to systematically crowd out other essential expenses. The validity of these cutoffs is most clearly demonstrated when aggregating them across dimensions. For instance, summing the “severely burdened” thresholds for housing (50 percent), food (30 percent), health care (17 percent), and child care (14 percent) yields a combined expenditure equivalent to 111 percent of a household’s post-tax and post-transfer income. This aggregate figure illustrates that a household experiencing simultaneous severe burdens mathematically has zero residual income available for other critical necessities, such as transportation, clothing, telecommunications, or emergency savings.

Similarly, adding the baseline “burdened” thresholds across these same four dimensions (30 percent, 18 percent, 10 percent, and 7 percent, respectively) consumes 65 percent of post-tax and post-transfer income. This leaves a highly constrained margin of just 35 percent of net resources to cover all remaining living expenses, demonstrating that households in this tier are forced to make continuous financial trade-offs to avoid material hardship.

Race and Ethnicity Definitions

We use the term “Latino” to refer to individuals of any race who identify as Hispanic or Latino in the ACS. All racial and ethnic categories constructed for this analysis are mutually exclusive. We classify individuals in one of five categories: white only, not Latino; Black only, not Latino; Latino (any race); Asian only, not Latino; or another race or multi-racial, not Latino. As a shorthand, we use the terms “white”, “Black”, “Latino” and “Asian” to refer to these groups.

In future analyses, we plan to disaggregate these data further to provide detailed insights into more groups, including American Indian and Alaska Native (AIAN) and Native Hawaiian and Pacific Islander (NHPI) populations. We also intend to use non-mutually exclusive racial categories to better account for multi-racial populations. Furthermore, we plan to disaggregate data for the Asian American population, as using a single category can obscure the needs and inequities within such a diverse group. Asian origin groups in the United States can vary widely in their economic status and education level.