Esri Updated Demographics methodology
Esri creates an Updated Demographics dataset for the United States. This topic describes the methodology used to create that data.
Tip:
Esri also provides a copy of this topic as a technical paper: 2026/2031 Esri Updated Demographics methodology (PDF).
Data vintage and variables
The 2026/2031 release of Esri's Updated Demographics introduces new estimates of school enrollment by grade level for the population aged 3 years and older. School enrollment estimates are available for the following educational categories: nursery or preschool, kindergarten, grades 1-4, grades 5-8, grades 9-12, college undergraduate, and graduate or professional school. These estimates extend the educational profile of the population and provide a consistent framework for analyzing enrollment patterns across early childhood, primary, secondary, and postsecondary education stages.
The 2026/2031 release also expands the resolution of household income estimates by age. In the prior 2025/2030 release, Esri increased the household income distribution from 9 to 20 income intervals, improving the granularity of income measurement. The current release builds on this foundation by aligning household income by age estimates with the same 20-interval income structure. As a result, household income by age increases from 63 cells (7 age groups by 9 income categories) to 140 cells (7 age groups by 20 income categories). This expanded matrix supports more detailed analysis of income distributions across life-cycle stages and improves consistency between income summary and cross-classified datasets.
Geographic coverage is further extended in the 2026/2031 Updated Demographics with the addition of three new summary geographic levels: Neighborhoods, ZIP Code Tabulation Areas (ZCTAs), and Combined Statistical Areas (CSAs). With these additions, the Updated Demographics product is now available for 30 geographic summary levels across the United States. This geographic expansion enhances flexibility for multiscale spatial analysis and supports a broader range of applications in planning, policy analysis, and market research.
|
Updated Demographics |
2026 |
2031 | |
|---|---|---|---|
|
Totals |
Total Population |
|
|
|
Daytime Population |
| ||
|
Time Series |
| ||
|
Households |
|
| |
|
Household Population |
|
| |
|
Family Population |
|
| |
|
Family Households |
|
| |
|
Group Quarters Population |
|
| |
|
Housing Units |
|
| |
|
Owner-Occupied Housing Units |
|
| |
|
Renter-Occupied Housing Units |
|
| |
|
Vacant Housing Units |
|
| |
|
Population Characteristics |
Population by Five-Year Age and Sex |
|
|
|
Population by Single-Year Age and Sex |
|
| |
|
Population by Age, Sex, and Race |
|
| |
|
Population by Age, Sex, and Hispanic Origin |
|
| |
|
Employed & Unemployed Population 16+ |
|
| |
|
Employed & Unemployed Population 16+ by Age Group |
| ||
|
Employed & Unemployed Population 16+ by Sex Group |
| ||
|
Employed & Unemployed Population 16+ by Race Group |
| ||
|
Employed Population 16+ by Industry |
| ||
|
Employed Population 16+ by Occupation |
| ||
|
Population 25+ by Educational Attainment |
| ||
|
Population 15+ by Marital Status |
| ||
|
Population by Generation |
|
| |
|
Housing Characteristics |
Households by Income |
|
|
|
Households by Income and Age of Householder |
|
| |
|
Households by Disposable Income |
| ||
|
Households by Disposable Income and Age of Householder |
| ||
|
Households by Net Worth |
| ||
|
Households by Net Worth and Age of Householder |
| ||
|
Owner-Occupied Housing Units by Home Value |
|
| |
|
Renter-Occupied Housing Units by Contract Rent |
|
| |
|
Households by Income Tiers |
|
| |
|
Indexes and Measures |
Diversity Index |
|
|
|
Housing Affordability Index |
| ||
|
Socioeconomic Status Index |
| ||
|
Percent of Income for Mortgage |
| ||
|
GINI Index of Income Inequality |
|
| |
|
Interdecile Ratios of Income Inequality |
|
| |
|
Share Ratios of Income Inequality |
|
| |
|
Wealth Index |
| ||
|
Age Dependency Ratio |
|
| |
|
Child Dependency Ratio |
|
| |
|
Senior Dependency Ratio |
|
| |
|
Economic Dependency Ratio |
| ||
|
Child Economic Dependency Ratio |
| ||
|
Working-Age Economic Dependency Ratio |
| ||
|
Senior Economic Dependency Ratio |
| ||
U.S. trends
From 2020 to 2026, the U.S. housing stock expanded nearly twice as fast as the population. Over this period, total housing units increased by 6.2 percent (from roughly 140.5 million to around 150 million units), while the population grew only 3.6 percent (from about 331.5 million to more than 340 million residents). As a result, there are now about 432 housing units per 1,000 people in 2026. This is up from about 424 per 1,000 in 2020—a modest but clear increase in housing availability relative to population.
Housing expansion outpaced population growth nationwide and in all four major regions with varying magnitudes. The regional patterns reflect pandemic-era migration shifts and differences in construction rates: Southern and Western regions saw robust population gains but even greater housing expansions, while the Northeast and Midwest experienced minimal population change yet continued housing additions, leading to larger relative housing surpluses.
In absolute terms, the South added the most residents (about 8 million) and housing units (about 4.3 million), reflecting strong Sun Belt growth. The West added approximately 2.9 million people and 2 million housing units. The Northeast and Midwest combined saw relatively small population increases (under 2 million combined) but still gained around 2 million housing units together, significantly raising their housing-per-capita ratios. All regions thus saw housing stock climb faster than population, increasing vacancy rates modestly.
From 2020 to 2026, the U.S. has experienced the slowest population compound annual growth rate of any decade prior to the 2020s. This reduced rate of growth is due to multiple factors including COVID-19 impacts, record-low immigration, and ongoing low birth rates. In contrast, housing growth has surpassed population growth due to low interest rates at the start of the decade, shifting housing preferences that accompanied remote work, and smaller household sizes. Multifamily construction has been particularly strong given smaller family sizes and the high cost of building and purchasing single-family homes.
Although housing growth has outpaced population growth, the U.S. has not solved its housing shortage problem. Given the reduction in household size, more housing units are needed per additional person or household (occupied housing unit), and the U.S. is still considered to have a cumulative housing shortfall from decades of underbuilding. All regions exhibit a common theme: households increased more than population but less than total units. This is because the U.S. average household size declined slightly from about 2.55 in 2020 to 2.49 in 2026. Any remaining excess units show up as vacancies: by early 2026, about 14.9 million housing units (10 percent of the housing stock) were vacant year-round, up from 13.7 million (9.7 percent) vacant units in 2020.
Summary totals
Forecasting change in the size and distribution of the household population begins at the county level with several sources of data. Esri incorporates intercensal time series and vintage-based county estimates from the U.S. Census Bureau. Because testing has revealed improvement in accuracy by using a variety of sources to track county population trends, Esri uses building permits, housing starts, and residential postal delivery counts. Beginning with this decade, Esri has modeled housing demolitions using data from the American Housing Survey (AHS). This view of housing is paired with a cohort survival approach to model changes to the population based on demographic composition. The result balances the measures of growth or decline from a variety of data series.
Measuring change in population or households at the county level is facilitated by the array of data reported for counties. Unfortunately, current data is not reported at the block group level. Past trends can be calculated from previous census counts. The American Community Survey (ACS) provides five-year averages. However, these sources are not recent. To measure current population change by block group, Esri models the change in households from multiple sources: Experian; the U.S. Postal Service (USPS); Zonda, a Hanley Wood company; and RealPage, in addition to several ancillary sources.
The USPS publishes monthly counts of residential deliveries for every U.S. postal carrier route. This represents the most comprehensive and current information available for small, subcounty geographic areas. Carrier routes are a fluid geographic construct that is redefined continuously to incorporate real changes in the housing inventory and occupancy plus administrative changes in staffing and budgets of local post offices.
Converting delivery statistics from postal carrier routes to census block groups is a complex challenge. Carrier routes are defined to deliver the mail, while block groups are constructed to collect and report census data. Comparing two areas that are defined for wholly different purposes provides one significant conversion issue. Carrier routes commonly overlap multiple block groups. In many cases, a carrier route encompasses disjointed areas that can be distant from each other, but block groups are rarely divided into multiple polygons. These overlaps require an effective method of allocating the postal delivery counts across multiple block groups.
Esri has developed a technique to link carrier routes to the correct block groups—using the actual locations of mail deliveries. Esri's proprietary Address Based Allocation (ABA) methodology was developed to solve the complex challenge of converting delivery counts from carrier routes to block groups. This allocation method assigns carrier routes using household addresses that are geocoded at the block level to serve as the foundation for the conversion. The approach is unbounded by geographic borders or arbitrary assumptions about the distribution of households or postal deliveries. ABA results have been tested extensively against decennial census counts, including an independent evaluation that involved data from four other vendors. This test confirmed the accuracy of Esri's ABA allocation method ¹. Another accuracy evaluation was conducted after the release of Census 2020 counts, highlighting improvements over the past decade and further validating the ABA technique ².
For more than a decade, Esri has licensed data from Zonda to track new residential construction for owned dwellings such as single-family homes and condominiums in the top U.S. housing markets. This database identifies the location and characteristics of individual construction projects, including total units planned, under construction, and closed by type of housing. This data is especially critical in tracking growth in previously unpopulated areas. Beginning with the 2016 updates, Esri has used an additional database from Zonda that more than doubles Esri's geographic coverage and the number of units planned and completed. The addition of this database gives the household and housing unit update a finer level of granularity and insight into smaller housing markets across the nation.
RealPage housing data is incorporated to capture the growing multifamily rental market. RealPage collects and maintains data on planned, new, and existing rental properties of multifamily and student apartments nationwide. This data source provides property-level characteristics such as the total number of units or beds, building type, number of stories, and occupancy, as well as market rent.
The best techniques are derived from a combination of models and data sources. Discrepant trends are checked extensively against independent sources and premium imagery data from Esri's ArcGIS Living Atlas. Finally, totals for block groups are controlled to the county totals.
Five-year projections
Projections are necessarily derived from current events and past trends. The past and present are known; the future must be extrapolated from this knowledge base. Even though projections represent the unknown, they are not uninformed. Pipeline projects slated for the future, as well as developments currently under construction, give Esri a unique view of the future landscape. Future residential construction information is integrated with a cohort component approach to model expected population change based on demographic characteristics. Guidelines for the development of projections also inform the use of those projections.
The recent past provides a reasonable clue to the course of future events, especially if that information is tempered with a historical perspective.
A stable rate of growth is easier to anticipate than rapid growth or decline.
The damaging effects of natural disasters cannot be anticipated. Esri makes every effort to assess the impact of sudden, catastrophic events such as strong storms, flooding, or wildfires.
The risk inherent in forecasting is inversely related to the size of an area: the smaller the area, the greater the risk.
The risk increases with the length of the projection interval. Any deviation of the projected trends from actual events is amplified over time.
Esri revises its forecasts annually to draw on the latest data. Projections can be enhanced with personal knowledge of an area to provide qualitative, anecdotal detail that is not captured in a national database. It is incumbent on the data user and the producers to incorporate as much information as possible when assessing local trends, especially for areas that are subject to boom-bust cycles or natural disasters.
Population and household characteristics
Esri incorporates a variety of data sources to update small areas such as block groups, beginning with the latest base and then adding a mixture of administrative records and private sources to capture change to the base. Shifting the base every year to the latest release of ACS data incorporates real change with sampling error. To establish a more stable base, Esri has built estimate bases for key variables such as income, labor force, and home value. The estimate bases combine the best data from ACS with other sources and allow better measures of change than are possible with ACS data alone. Periodic changes to the estimate bases are necessary to collect current changes. Base changes impact comparability of the annual data but provide more reliable estimates. Demographic updates must incorporate both traditional and new data sources to remain current.
The population by age and sex is projected using a cohort survival model that separately calculates the components of population change by age and sex. Applying survival rates specific to the cohort carries a 2020 population base forward. Changes in the population by age and sex diverge at the household level. For example, an area that is losing population can age more rapidly with the loss of population in prime migrant ages, 20-34 years—unless there is a college nearby. Neighborhoods near colleges sustain high turnover from student populations but retain their youthful age distributions.
To capture these variations, Esri's model first separates the group quarters population from the household population and keys the calculations to the size and characteristics of the population. This stratification identifies several patterns of change by age and sex that can be applied in a cohort survival model.
The changing profile of the U.S. population requires measuring population change by race and Hispanic origin. The American identity is shaped by diversity. Tracking the changing patterns of race and ethnicity provides a current portrait of our society. Historical trends in race and ethnicity combined with the most current data sources by race and Hispanic origin, including population estimates by county and state from the Census Bureau and survey data from the ACS, are analyzed to establish county population by race and Hispanic origin. Forecasts by block group combine local changes in the distributions by race and projected changes for counties. The last step controls block group distributions to county totals by race and Hispanic origin.
The changing face of our nation is evident in Esri's Diversity Index, which summarizes racial and ethnic diversity in an area. Esri's definition of diversity is two-dimensional and combines racial diversity with ethnic diversity. This measure shows the likelihood that two people, chosen at random from the same area, belong to different races or ethnic groups. In theory, the index ranges from 0 (no diversity) to 100 (complete diversity). An area's diversity index tends toward 100 when the population is more evenly divided across race and ethnic groups. If an area's entire population is divided evenly into two race groups and one ethnic group, the diversity index equals 50. As more race groups are evenly represented in the population, the diversity index increases. Race and Hispanic origin data is reported by the Census Bureau and other agencies as grouped summary data; therefore, in practice, the diversity index will not reach the maximum value of 100. Nationally, Esri's Diversity Index has risen from 71 in 2021 to 73 in 2026, with a forecast to 74.6 in five years.
Diversity also describes the composition of American households. Esri uses the Census Bureau's definition of families and family households. Families include a householder and one or more people living in the same household who are related to the householder by birth, marriage, or adoption; therefore, family households are equal to the number of families. Family households can also include unrelated nonfamily members. Family households are modeled from Census 2020, Current Population Survey (CPS), and ACS data. Unlike the previous decennial census, Census 2020 provides counts of the population in families only at the state level. This information from the 2020 Supplemental Demographic and Housing Characteristics File is incorporated into the workflow, but Esri's approach to modeling this information also takes advantage of ACS data available at smaller geographic levels. Average family size sits at 3.11 for 2026.
The attendant change in average household size has shown a decline from 2.58 in 2010 to 2.55 in 2021 and a continued fall to 2.50 for 2026. Average household size is used when forecasting the change in household population from the change in households. Average household size is traditionally one of the most predictable components of the forecasts and serves as a link between the population and household universes. Household forecasts are predicated on local patterns of change, which are controlled to more constant trends for states and counties.
Few block groups represent a cross section of U.S. households. For example, in areas that gain population from immigration, the trend in average household size is an increase. To distinguish local variation, Esri's model is keyed to the characteristics of households at the block group level. This stratification identifies several patterns of change by household type that are applied to forecast trends in the characteristics of households—both family composition and tenure. Local change is emphasized in the 2026/2031 forecasts of households for counties and block groups. National and state trends are monitored with sources such as the CPS and ACS from the Census Bureau and then applied as controls.
A mixed source model approach is used to forecast 2026 educational attainment and marital status, combining higher-level and timelier 1-year ACS data with 5-year lower-level ACS data as well as national statistics from the CPS. Adjustments are factored for changes to the base population's characteristics. Forecasted distributions are applied to Esri's 2026 population aged 15 years and older to update marital status. Similarly, educational attainment is updated for the population aged 25 years and older.
Lastly, a multi-source model approach is also used to estimate 2026 school enrollment, combining detailed age distributions with enrollment patterns derived from ACS Public Use Microdata Sample (PUMS) data, and neighborhood context from ArcGIS Tapestry Segmentation. Estimated distributions are applied to Esri's 2026 single-year-of-age population at the block group level to produce enrollment estimates by grade level.
Housing data
Esri's housing updates include total housing units, occupancy, tenure, home value, and contract rent. Total housing unit updates are created from recorded changes in the housing inventory and estimated changes in occupancy rates since April 2020, applied to Census 2020 base data. Recorded change in the housing inventory is culled from several data sources, including multiple construction data inputs from Zonda and RealPage, data for new manufactured homes placed by state from the Census Bureau, and building permits for permit-issuing places and counties. Numerous independent sources are used to obtain detailed information on housing development data where no building permits exist. Independent estimates of change in occupancy are calculated from USPS residential lists, the ACS, and various state and local data sources. Additionally, data from the CPS and the Housing Vacancy Survey (HVS) from the Census Bureau is used to model trends in occupancy.
Data for tenure represents owner- and renter-occupied housing units. Together, the two components sum to total households or total occupied housing units. Esri's tenure estimates and projections incorporate the updated demographic profiles released in the Census Bureau's DHC product and new residential construction data. Moreover, a time series model based on data from the HVS, combined with changes in the CPS, the ACS, and intercensal data, guides tenure forecasts. With a blend of top-down and bottom-up techniques, the forecasts take advantage of the latest information from survey data at higher levels of geography while employing local characteristics at the lower levels. The small-area models use more geographically granular trends from ACS as well as integrate the Zonda and RealPage housing data to update an area's tenure profile. ACS tenure time series data used in the models was smoothed to further reduce survey noise by tempering outlying data points. Data from lower levels of geography is controlled to higher levels to produce tenure updates. Changes in owner versus renter occupancy are forecasted independently and then controlled to total households.
Esri reports home value for owner-occupied housing units. A total of 26 home value intervals are reported. Information is collapsed to a 13-interval distribution as well for comparison with previous releases. Summary measures of home value include medians and averages that are calculated from the distributions of home value. Medians represent the middle of the distribution or the point that splits the distribution equally ³. Medians are calculated using linear interpolation unless the median falls in the highest (>$2,000,000) interval. Following the Census Bureau's convention, this median is reported as $2,000,001 because housing value in the upper interval is top-coded to $2,000,000. Due to limited data availability for these high-valued homes, Esri top-codes average home value to $2,250,000.
Esri tracks the change in home value using several sources, including annual estimates from the ACS, the Home Price Expectations Survey from Pulsenomics, and the House Price Index (HPI) from the Federal Housing Finance Agency (FHFA). The Home Price Expectations Survey relies on a survey of more than 100 industry experts to forecast growth in the housing market. This source is a key input to Esri's forecasts. The HPI is designed to monitor changes in average home prices based on repeat sales or refinancing of the same properties. The index is derived from mortgage loans purchased or securitized by the Federal National Mortgage Association (Fannie Mae) or the Federal Home Loan Mortgage Corporation (Freddie Mac).
Esri's 2026 home value estimates continue to show growth from the previous year, though home values are increasing at a much slower rate than they were during the pandemic years. Still, home values remain at an all-time high, and the previously expected (by many experts) downturn in the housing market has not yet materialized. There continues to be great uncertainty in the housing market, particularly in terms of where housing supply and mortgage rates will trend in the coming years. Esri shows a median home value of $376,272 for 2026 with a projected 3.9 percent annual change between 2026 and 2031.
Esri's model emphasizes the importance of a stable forecast base. Employing both the ACS's historical 5-year estimates 4 and household survey data, Esri's 2026 estimates begin with an updated forecast base that uses the growing stability of ACS data. Once every few years and particularly during real estate market cycles, it is prudent to reset the base to capture the current housing landscape. Though this does preclude comparisons to past updates, especially for small areas, the base provides a strong foundation to measure change. Local estimates of home value change incorporate supply-demand characteristics, the socioeconomic traits of householders in the area, and trends assessed for larger markets.
Esri uses current housing and income data to provide a snapshot of affordability. Esri's approach to measuring housing affordability 5 uses an index to quantify the ability of a typical resident to purchase an existing home in an area. Employing information from a variety of sources to estimate the national average contract mortgage rate, an interest rate of 6.2 percent is estimated for Esri's 2026 Housing Affordability Index model. A 30-year mortgage is assumed, with a down payment of 20 percent of the home price. Property tax and homeowners insurance rates are determined from the latest ACS 5-year data, and Esri's model follows the Federal Housing Administration's guidelines for debt service ratios. Additionally, the 2026 Percent of Income for Mortgage (POIFM) quantifies the percentage of median household income dedicated to mortgage payments on a home priced at the median value.
Esri reports contract rent for renter-occupied housing units at 25 intervals as well, using a similar modeling methodology to that of home value. Information is collapsed to a 13-interval distribution as well for comparison with previous releases. Contract rent is defined as the monthly rent agreed to or contracted for, apart from any utilities, fees, or other services that may be included. With the contract rent distribution, the 25th interval represents an estimate of renter households that do not pay rent. This category may include units that are provided to compensate caretakers, ministers, and others who tend to live at their place of employment. Additionally, this category can include units that are owned by relatives or friends and allow occupancy without charging rent.
Summary measures of contract rent include medians and averages that are calculated from the first 24 intervals of the contract rent distribution. If the median falls in the highest interval (>$3,500), the median is reported as $3,501. Average contract rent is top-coded to $3,750. Esri shows a median contract rent of $1,444 for 2026 with a projected 2.9 percent annual change between 2026 and 2031.
Labor force data
Esri forecasts the 2026 employed and unemployed population aged 16 years and older who are either working or actively looking for work, also known as the civilian labor force. U.S. gross domestic product (GDP) declined from the previous quarter in Q1 2025 but improved over the remainder of the year. However, the most recent estimate shows GDP growth slowed to an annual rate of 0.5 percent in Q4 2025. Under current economic conditions, the total U.S. workforce exceeds 168 million people.
The civilian labor force does not include active-duty military, institutionalized individuals, and those not actively looking for work. The employed population 16 years and older is further broken down by industry and occupation. Esri also provides additional labor market detail in estimates of employment and unemployment by four age groups (16 to 24, 25 to 54, 55 to 64, and 65 and older) as well as breakouts by sex and race groups.
Estimates of the civilian labor force are modeled using 1-year and 5-year ACS employment tables and CPS information from the U.S. Census Bureau as well as data from the Local Area Unemployment Statistics (LAUS), Occupational Employment Statistics (OES), and Current Employment Statistics (CES) programs of the Bureau of Labor Statistics (BLS). Federal statistical surveys are the principal sources for labor force trends. The 2026 employment and unemployment estimates are developed from a block group base constructed from 1-year and 5-year ACS labor force tables and current sources. Esri's updated employment by industry and occupation estimates capture temporal change from multiple federal statistical sources: the ACS and CPS from the Census Bureau and the CES and OES programs from the BLS. In 2026, the ACS-derived base was updated to leverage the most recent survey data, producing more current labor force profiles for small areas. The methodology used to estimate the labor force by age, sex, and race was also enhanced to better preserve relationships among these characteristics. As a result, comparisons with Esri labor force estimates released prior to 2026 are not recommended.
Esri's Socioeconomic Status Index (SEI) quantifies disparities in social position of an area's inhabitants and is built from a broad range of demographic, housing, and socioeconomic inputs. This index ranges from zero (lowest relative status) to 100 (highest relative status). More information about this measure can be found in Esri's SEI data guide 6.
Household income
Esri's 2026 household income estimates are reported in nominal dollars for households as of July 1, 2026. Esri's household income and related estimates reflect calendar year 2026. Similarly, forecast year 2031 estimates are reported for forecasted households as of July 1, 2031, and represent household income for calendar year 2031. Esri's estimates of household income are benchmarked to the latest ACS data (2024) from the Census Bureau; therefore, change between 2023 and 2025 is gauged in the model. Accounting for current inflation and historical income change, Esri estimates 29 percent annualized growth in median household income since 2024. The 2026 median household income stands at $85,893, with average household income at $122,692.
Household income distributions are estimated for areas with 10 or more households. Esri adheres to the Census Bureau's definition of money income, which includes various sources of income for persons 15 years and older such as earnings, unemployment compensation, Social Security, Supplemental Security Income, public assistance, veterans' payments, survivor benefits, disability benefits, pension or retirement income, interest, dividends, rent, royalties, estates and trusts, educational assistance, alimony, child support, and other financial assistance. Notably, money income excludes capital gains and deductions for personal taxes, Social Security payments, union dues, and Medicare.
It is important to recognize the differences between the Census Bureau's household income and other sources of income, including the Bureau of Economic Analysis' (BEA) personal income estimates. BEA calculates personal income as part of its mission to produce national income accounting estimates such as gross national product. The Census Bureau collects income statistics to satisfy its objective to enumerate and describe the population of the United States. Caution should be exercised when comparing income estimates from different sources, as variations in definitions, data collection methods, reference areas, and population coverage can lead to discrepancies 7.
Starting in 2025, Esri expanded the income categories from 9 intervals to 20 intervals. This expansion includes the full detail provided by ACS plus additional detail for households earning more than $200,000, which are now distributed into five distinct categories. The new intervals are as follows:
| <$10k | $30k - $35k | $60k - $75k | $200k - $250k |
| $10k - $15k | $35k - $40k | $75k - $100k | $250k - $300k |
| $15k - $20k | $40k - $45k | $100k - $125k | $300k - $400k |
| $20k - $25k | $45k - $50k | $125k - $150k | $400k - $500k |
| $25k - $30k | $50k - $60k | $150k - $200k | $500k+ |
Esri continues to provide the summary categories previously reported, including the following:
| <$15k | $75k - $100k |
| $15k - $25k | $100k - $150k |
| $25k - $35k | $150k - $200k |
| $35k - $50k | $200k+ |
| $50k - $75k |
To estimate income for households, Esri evaluates an extensive list of sources including both federal and proprietary sources. The review of national surveys includes the ACS (both 1-year and 5-year estimates), the BEA's local personal income series, the CPS, and the BLS's Consumer Price Index. ACS Public Use Microdata Sample (PUMS) data is relied on to model household income above $200,000.
Estimates for household income are in nominal dollars. In other words, the growth of income attributed to inflation is included in the estimate. Esri models nominal household income directly. With inflation estimates only available at the national and regional levels and for selected major cities, and a lack of local area inflation data, estimating local real household income is imprudent. Esri tracks national inflation rates to guide both current-year and forecast-year estimates. Expected national inflation is based on trends from 5- and 10-year break-even rates. These rates are computed from the spread between nominal and inflation-adjusted Treasury securities as of the end of November 2025. Break-even rates represent an estimate of the average expected inflation premium that market participants are pricing into these securities over the two time horizons. The annual inflation factor is forecasted at 2.3 percent.
The estimation of households by extended income categories leverages the details available in ACS PUMS data, which provides a weighted subsample of ACS responses. Since PUMS data is only available for Public Use Microdata Areas (PUMAs), Esri employs weighted geographic apportionment for block group level analysis. Esri's income estimates are built on an annually updated forecast base, capitalizing on historical ACS 5-year estimates 8. Both sampling and nonsampling errors contribute to the instability of time series data for small areas, prompting Esri to design parameters that quantify and normalize this instability, which results in a robust base for measuring income changes.
After forecasting the state income distributions, household income is estimated for block groups. Esri's income forecasts are uniquely designed to distinguish local variation, changes in income inequality, and urbanicity as differentiators of income growth. The model correlates the characteristics of households at the block group level with changes in income. This stratification identifies several patterns of change by household type that are applied to forecast trends in income. Modeling links the current income change to all households with similar socioeconomic characteristics.
Areas with small household bases or missing base data, where the model is unable to capture the local variation, are forecast with another level of modeling to capture the change in income by strata (a group of areas classified by their sociodemographic characteristics). Separate forecasts of the change in income by strata are aggregated to compose the income distributions .
Summary measures of household income include medians and averages that are calculated from the distributions of income. A median represents the middle of the income distribution or the point that splits the distribution equally. A median is calculated from the income intervals of the distribution using Pareto interpolation, unless the median falls in the lowest (<$10,000) or highest (>$500,000) interval. For the lowest interval, linear interpolation is used. When the median falls in the upper interval, it is reported as $500,001 because households in the upper interval are top-coded to $500,000. The shift to reporting 20 intervals impacts medians most significantly in areas with a high concentration of income at the tails of the distribution.
Averages are computed from estimates of aggregate income, using a midpoint approach which assigns the same dollar income value, or an average income, to each household in an interval. This is straightforward for bounded intervals, but for the open-ended top, interval modeling is required to determine average household income. Whereas in previous years, midpoints were designed by broad sociodemographic groupings, new methods introduced in 2025 create local area midpoints to better align with historical ACS average income.
It should be noted that a change in the number of intervals has implications for summary statistics including income tiers and income inequality measures, as well as for other variables that build from the household income data such as age by income, disposable income, and net worth tables.
All household income summary statistics are computed using the detailed 20-interval income distribution. Esri's methods to compute median income rely on Pareto interpolation within the interval in which the median falls and are therefore sensitive to the number of intervals. In other words, median income calculations on a 9-interval versus 20-interval income distribution will yield different results if the median falls in a different income range. Average income calculations are influenced by the definition of intervals because the distribution used for aggregate income changes. The most significant impact of the new interval structure, due to the top-coding shift from $200k to $500k, is in high-income areas, which are now displayed with more detail.
Esri has made available additional summary measures to quantify income inequality. Computations of the Gini Index, interdecile and share ratios of income inequality, as well as households by income tier, are based on the annually updated 20-interval household income. The shift to reporting detailed and extended income intervals and the recalibration of the average income model does influence all summary measures, but positional measures, such as interdecile ranges, are more stable with less exposure to the tails of the distribution. For further information, refer to Esri's 2025/2030 Income Tiers and Measures of Income Inequality Methodology Statement 9.
Household income by age of householder
Household income by age of householder is reported for seven age of householder groups with 20 income intervals. Starting with this update, Esri is providing 20 income intervals for each age group consistent with household income estimates.
To estimate age of householder by household income for 20 income brackets with detail extending beyond $200,000, it was necessary to use historical data from the ACS PUMS in combination with ACS five-year data to build a block group level forecast base. This is a break from the past since Esri historically relied on a combination of public and private data sources to capture more detailed age breaks that the ACS does not provide. The new model leverages PUMS data to break out the four broad age groups provided in the ACS age by income tables into 7 age groups consistent with Esri's age by income releases. With multiple layers of change, caution should be exercised when comparing current estimates with previous years.
To update the age distribution of householders, the ratio of householders by age to the population by age in Census 2020 is applied to the 2026/2031 population by age estimates, taking into account the change in group quarters. Base distributions of household income by age of householder at the block group level are then fitted to Esri target distributions of households by income and age of householder.
The methods for median and average calculations follow those used for household income. Midpoints designed in the household income model are applied to each age group, ensuring an exact summation of income by age cells to total aggregate income.
Disposable income
Disposable income is estimated in nominal dollars for the nine summary income intervals and seven age of householder groups. Data sources to evaluate detailed disposable income are limited due to top-coding in both before-tax and after-tax variables in our sources, and therefore, Esri will continue to estimate nine income intervals by age consistent with prior updates.
Disposable income represents money income after taxes—an estimate of a household's purchasing power. The proportion of household income left after taxes is estimated from special studies conducted by the Census Bureau to simulate household taxes. Esri's 2026 disposable income estimates incorporate data from the 2025 Annual Social and Economic Supplement of the Current Population Survey (ASEC).
Four types of taxes are deducted: federal individual income taxes, state individual income taxes, FICA (Social Security), and federal retirement payroll taxes. Property taxes for owner-occupied housing are no longer available in the ASEC and therefore dropped from Esri's model. Internal Revenue Service tax rates are used as guidelines for model testing. Esri then applies the proportions of after-tax earnings to income intervals that are cross tabulated by age of householder for each state. State-specific proportions account for the variation in taxes by state. The proportions, or multipliers, are then applied to the age by income forecasts for block groups and counties to calculate disposable income.
The methods for median disposable income calculations follow those used for household income by the age of householder. Beginning in 2025, average disposable income calculations introduced new methods to estimate disposable income midpoints that better leverage local patterns, rather than trends derived for broader socioeconomic groupings. Estimation starts with the 9-interval midpoints derived from the 20-interval household income midpoints. Then, after-tax proportions of income stratified by age and state are applied to derive disposable income midpoints. Average disposable income estimates now benefit from modeling at a more granular level, as well as midpoints influenced by the 20-interval detail, which most significantly impacts top-heavy or bottom-heavy income distributions. This is reflected by a broader range of average disposable incomes by age (lower lows and higher highs) in small areas.
Net worth
Current income is only one component of a household's financial security. Householders' net worth or accumulated wealth reflects their ability to stay afloat during a financial downturn as well as save for future retirement. Net worth is estimated with data on household wealth that is collected from the Survey of Consumer Finances (SCF) from the Federal Reserve Board from 1992 through 2022. These triennial surveys feature enhanced representation of wealthy households through the comprehensive measurement of net worth components. By definition, net worth equals total household assets less any debts, secured or unsecured. Assets include ownership of homes, rental properties, businesses, individual retirement accounts (IRAs) and Keogh accounts, pension plans, stocks, mutual funds, and motor vehicles. Examples of secured debt include home mortgages and vehicle loans; unsecured debt includes credit cards and other bills or certain bank loans.
The 2026 update of net worth builds upon the 2022 SCF and incorporates more recent trends in household net worth from the Federal Reserve's report of the Financial Accounts of the United States. Between 2019 and 2022, the SCF measured an increase in median net worth of 37 percent and an increase in average net worth of 23 percent. Esri's current median net worth estimate is $276,821 and builds in a 9.5 percent annualized growth in the three years since the last SCF data point.
Beginning in 2019, Esri's total net worth is reported for 12 intervals to include an upper interval of greater than $2 million. Net worth is also reported for the seven age of householder groups, by 10 net worth intervals. Summary measures of net worth include medians and averages, which are calculated from the distributions of net worth. Similar to household income methods, a median is calculated from the net worth intervals of the distribution using Pareto interpolation, unless the median falls in the lowest (<$15,000) or highest interval. For the lowest interval, linear interpolation is used. When the median falls in the upper interval, it is reported as $1,000,001 for net worth by age of householder and $2,000,001 for total net worth.
Averages are computed from estimates of aggregate income, using a midpoint approach which assigns the same dollar income value, or an average income, to each household in an interval. Methods to compute the midpoints changed in 2026. In previous updates, the midpoints for the first and last interval were modeled while the true midpoint was employed for all other intervals. While continuing to design midpoints by sociodemographic and economic strata, the new midpoints are linked to the SCF microdata rather than aligning net worth midpoints with household income midpoints. Midpoints are now modeled for every net worth bracket, which means that for bounded intervals, the midpoint is no longer a true middle value, but reflects patterns revealed in the survey. This approach significantly improves the estimation of average net worth while maintaining national trends.
The foundation of the net worth data is Esri's age by household income and tenure estimates, with the relationship between net worth and sociodemographic characteristics captured from historical SCF data. The new detailed estimates of age of householder by income are leveraged in the 2026 net worth model; this improvement most significantly impacts areas with a high concentration of households in the lowest or highest net worth brackets .
Esri's Wealth Index is compiled from a number of indicators of affluence, including average household income and average net worth. The concept of wealth is defined by more than above-average household income. Wealth also includes the value of material possessions and resources. Esri captures both income and the accumulation of substantial wealth or the abundance of possessions and resources in its identification of the wealthiest areas in the country. The index represents the wealth of the area relative to the national level. Values exceeding 100 represent above-average wealth.
Geography
Current-year estimates and forecasts are prepared initially for counties and block groups which are then allocated to census blocks. Allocation of block group estimates to blocks maintains demographic consistency across all variables, which allows for data at block-based geographies to be consistent as well.
For all other user-defined boundaries such as rings or drive-time polygons, block weights are applied to block group data to apportion demographics within these areas.
Changes in the geographic areas for which data is tabulated and reported are critical to the analysis of trends. Esri reports data for 30 geographic schemas in the United States and 19 in Puerto Rico, a significant increase from last year's update. These include many schemas from Census Bureau/TIGER (Topologically Integrated Geographic Encoding and Referencing) as well as vendor boundaries and Uber H3 hexagons. New geographic schemas for which current-year estimates and forecasts are available starting with the 2026/2031 updates include the following:
For a complete list of available geographies, see the Available geographies inventory in Esri's United States topic. The provision of small-area data in Esri software also allows users to define their own areas of interest 10.
Data is reported in 2020-based geography (TIGER 2025) for most of the standard legal/administrative and statistical areas. Legal/administrative areas are typically those that involve government officials and the administration of services or representation. These are usually official government boundaries that would exist without the need to present statistical data. In general, the Census Bureau accepts boundaries that are provided by official local entities. Statistical areas do not require official documentation and do not typically involve government officials administering services. Statistical areas are established for purposes of data tabulation and presentation purposes as well as research.
The 2026/2031 updates reflect the metropolitan and micropolitan statistical areas released by the U.S. Office of Management and Budget (OMB) in July 2023. There are 387 metropolitan and 538 micropolitan areas. Congressional districts represent the 119th Congress. The place inventory is from TIGER 2025 and contains 32,058 areas.
ZIP codes, which are defined solely by the USPS to expedite mail delivery, can change monthly or whenever the USPS revises delivery routes. ZIP codes do not represent standard census geographic areas for data reporting. ZIP code boundaries are not contiguous with census geographic areas or stable over time. Data estimated for ZIP codes is also subject to change. Residential ZIP code data is estimated from block data established from block group estimates, using a correspondence created by assigning census block points to ZIP code boundaries from TomTom. The vintage of the ZIP code boundaries is second quarter of 2025; the total number of residential ZIP codes in this release is 32,184.
The integration of demographic and spatial analysis has not only enabled the development of more accurate block group totals, it has also provided the opportunity to update block totals. Blocks are the lowest common denominator in the geographic hierarchy and progress to block groups, tracts, counties, and states. Blocks are most useful in the estimation of data for polygons, which can be any area outside the geographic hierarchy, from retail trade areas to user-defined polygons (including circles and drive-time polygons). For most areas, the application provides a good estimate for the polygon. If the relationship between the underlying blocks and the parent block groups has changed significantly since 2020, the estimate cannot incorporate that change unless both blocks and block groups are updated.
References
See the following references for additional information:
Esri Vendor Accuracy Study: 2010 Estimates versus Census 2010
Due to the delayed release of ACS 5-year 2024 data, Esri's home value base was not updated for the 2026 model.
Comparability of Current Population Survey Income Data with Other Data
Due to the delayed release of ACS 5-year 2024 data, Esri's household income base was not updated for the 2026 model.
Esri's Income Tiers and Measures of Income Inequality Methodology
Understand Esri's U.S. geography and methods used to estimate data for any user-defined polygons