General
Data recency
All information shown in the Native Economic Trends (NET) visualization reflects data available as of September 2025. The Center for Indian Country Development (CICD) plans to update this tool on a roughly annual basis.
Programming acknowledgments
All compilation, processing, and calculations are performed in R. CICD is particularly grateful to the authors of the ipumsr, tigris, and tidycensus packages and their dependencies, which we use for all interactions with U.S. Census Bureau (Census) data used in this project.
Data providers
Census’ Application Programming Interface (API) services provide ready programmatic access to Census’ public data. We access all American Community Survey (ACS) data for this project via API and gratefully acknowledge these services.
Decennial census data prior to 2000 are not yet available via Census API at this time. For consistency, we obtained all decennial census data for this project from the Integrated Public Use Microdata Series (IPUMS) National Historical Geographic Information System (NHGIS), a publicly available academic resource housed at the University of Minnesota for which we are also appreciative.
To ensure alignment between numeric data and the geographic boundaries in effect at the time of collection, we use shapefiles from the same source and reference year as the data; that is, NHGIS-provided TIGER/Line-derived shapefiles for decennial values and Census-provided TIGER/Line shapefiles for ACS values. The exceptions to this pattern are geography names for 2000 that were missing from the NHGIS-provided shapefile and taken instead from Census.
Linking data over time
Geographies
The primary geographic units in NET are Census’ American Indian/Alaska Native/Native Hawaiian Area geographies. We refer to these as Native geographies. CICD acknowledges that Native geographies tabulated by Census and displayed in this tool do not encompass all places having current or historical associations with Native nations.
The system of geographic identifiers used for Native geographies in Census products has changed over time, most significantly between 1990 and 2000. This system relies on political status in its encoding, so the identifier for a tribally associated geography changes should the tribe’s political status change. Refer to Census documentation for the Native geography code categories in effect for each NET data year with the exception of 1990, which is available via IPUMS NHGIS.
An additional factor of note is that Native geographies commonly change shape and expand—sometimes considerably—through land transfer and acquisition transactions. Merges and splits in geographies from one Census collection year to another also occur.
Where programmatically linking geographies over time is not possible, we link them manually through spatial analysis. Some geographies are not observable in all years. In NET, no Native geographies observed over the data period are excluded.
Conceptually, geographies are grouped by category and name in the visualization using their most recently observed record. Geographies that were split or combined over time are shown as one line but remain distinct in supporting information visible when mousing over a datapoint. For example, the four 1990 geographies that were later combined into Navajo Nation are included in the collective Navajo totals, but when mousing over the line at 1990, their separate names are shown. A geography’s name in a given data year is displayed in NET as it was documented by Census at that time.
State and U.S. boundaries reflect the most recent data year (currently 2023) and are assumed to have remained the same over the NET period. States’ identifiers are their FIPS, or Federal Information Processing Standard, codes.
Census survey instruments and reporting
The decennial census occurs once each decade and aims to count the entire U.S. population. In addition to core questions asked of every person using a “short-form” questionnaire, decennial censuses in 2000 and before included a “long-form” version that was administered to a subset of households to collect detailed socioeconomic data. To CICD’s knowledge, 1990 is the first year for which publicly released, standardized decennial census data are tabulated for Native geographies nationwide. This is why NET’s data series begin in 1990.
The ACS replaced the long-form decennial census in 2005. The ACS asks questions of weighted samples of residents rather than the full population. Thus, data from the ACS—like data from the long-form decennial census—are estimates, rather than directly measured quantities. Wherever the option exists for a given decennial measure shown in NET, we use short-form (that is, full-count) rather than long-form data.
ACS data are available for more populous Native geographies in one-year increments, but only in five-year increments for all Native geographies (that is, the first available is 2005–2009). These five-year ACS estimates represent averages over a given time period. This means that overlapping ACS time periods represent, in effect, rolling averages. In NET, we use the most recent ACS five-year estimates (currently 2019–2023) as the end of the series and work backward, showing nonoverlapping five-year periods. ACS data are displayed as the midpoint of the five-year period; for example, 2019–2023 is shown as 2021.
In addition to shifts in survey instruments in recent decades, Census has also made periodic adjustments to questionnaire layout, question wording, and response processing. The impacts of new privacy protection measures are an area of active research.
For completeness, NET includes both sets of data: decennial and ACS. Because they reflect different methodologies, they are shown as separate series. NET users can choose to exclude one or the other with an optional filter. In future updates to the tool, decennial datapoints will stay in place, while five-year ACS datapoints will shift forward to present the most up-to-date data available.
Racial identification in Census survey data
Information about race shown in NET was self-reported by survey respondents.
Individuals who report a single race and identify as American Indian or Alaska Native (AIAN) alone are included here in the category denoted AIANa. Individuals who report a single race and identify as Native Hawaiian and Other Pacific Islander (NHOPI) alone are included in the category denoted NHOPIa. (As is evident in the name, this broad category represents populations other than Native Hawaiians, including both Indigenous and non-Indigenous groups.)
Beginning with the 2000 decennial census, individuals could self-identify as more than one race. Respondents who self-identify as AIAN or NHOPI alone or in combination with one or more other races are included here in the racial categories denoted AIANac and NHOPIac, respectively. Wherever possible, we use the more inclusive NHOPIac and AIANac definitions—especially important given that well over half of Native census respondents self-identify as more than one race. However, for many measures, AIANa and NHOPIa are the only categorizations publicly available.
Over the years, Census’ race and ethnicity question structures and classification approaches have changed. Researchers have documented the sizable impacts of these changes on the comparability of demographic estimates over time, particularly among Native-identifying individuals.
Two differences in decennial census data from 1990 bear special mention. In publicly available data, individuals who would come to be classified as NHOPI were included in a broader Asian or Pacific Islander reporting category. As a result, we do not include NHOPI-specific measures for 1990. In addition, 1990 data shown in NET as AIANa reflect a reporting category of American Indian, Eskimo, or Aleut.
A place-based resource
Census reporting is fundamentally place-based, not tribe-based. AIAN and NHOPI individuals who reside within the boundaries of a tabulated Native geography are counted in data for that geography. AIAN and NHOPI individuals with community ties to that geography—but who reside outside its boundaries—are not included in its Census data.
Other data limitations
Census surveys are the primary nationwide source of publicly available data about Native communities. However, these data can have significant limitations—particularly in geographies that are remote, are small in size, or have low populations. Undercounts and large margins of error for sample-based estimates are known data-quality challenges.
The data download that accompanies NET includes a population field. Users can use this field to filter geographies for analysis based on total population.
For further discussion of the nuances of Census data efforts related to Native communities, see Akee et al. 2014 and 2025.
Additions and corrections
While we have made strong efforts to ensure the accuracy of manually compiled information, CICD welcomes additions and corrections from representatives of tribes, federal government agencies, and other stakeholder bodies. Please email cicd.data@mpls.frb.org to share input.
Compilation details
Measures from the American Community Survey
Unless otherwise noted below, all ACS-derived measures follow the conventions, definitions, and sourcing described in the About the Data documentation for Native Community Data Profiles.
Measures from the decennial census
Decennial-derived measures are described below. For full transparency, IPUMS NHGIS table specifications are given as [dataset group code].[source table code].
All decennial census dataset groups from 2010 and 2020 represent full-count data—that is, questions that are asked of all people and about every housing unit. Dataset groups from 1990 and 2000 that begin with SF1, STF1, SF2, or STF2 are also full-count data. In contrast, those that begin with SF3, STF3, SF4, or STF4 are sample-based data collected from subsets of the population.
IPUMS NHGIS tables are often unaltered from those published by Census. In other cases, a suffix indicates that a table does not directly correspond to a published table and has been adapted in some way to facilitate use by researchers (e.g., splitting a table to provide a breakdown by race).
For example, 1990_STF1.NP1 refers to table NP1 from NHGIS dataset 1990_STF1. These data were collected by Census via the 1990 short-form questionnaire, and NP1 matches the published Census source table P1. The data source 2000_STF2.NPCT001A refers to information collected from a sample of the population via the 2000 long-form questionnaire. These data are comparable to Census table PCT1 with a race and ethnicity breakdown added.
For further details about IPUMS NHGIS structure and naming conventions, please see the NHGIS datasets overview page.
Inflation adjustment
Dollar measures are inflation-adjusted to the most recent data year (currently 2023).
Modes
The tool’s Compare mode allows users to select a single measure of interest for a single primary geography and show its trendline alongside those of one or more geographically relevant comparison geographies. For Native geographies, available state comparisons are those with which the Native geography has any amount of spatial overlap.
In the tool’s Combine mode, users can select any number of geographies or groups of geographies and up to two measures (such as total population and total workforce) to be plotted together. Values are calculated on-the-fly as if the selected geographies were combined into one unit. For example, when a user selects all “Federal reservations, off-reservation trust lands, and joint-use areas” in the tool, the data shown reflect all Native geographies with current Census geographic codes from 0001 to 4989, combined into one geographic unit. In Combine mode, measures that are sums (such as housing units) are simply added together. Measures that are percentages (e.g., labor force participation) or medians are weighted by the appropriate unit—described below—and then averaged. Note that correlation between two measures does not necessarily mean that one causes the other.
About the People data
Total population: We use decennial tables 1990_STF1.NP1, 2000_SF1a.NP001A, 2010_SF1a.P1, and 2020_DHCa.P1.
Native population: We use the total population values above as denominators. AIANa and NHOPIa measures are then calculated using decennial tables 1990_STF1.NP6, 2000_SF1a.NPCT007A, 2010_SF1a.P3, and 2020_DHCa.P3. For AIANac measures, we use 2000_SF2.NPCT001A, 2010_SF1b.PCT4, and 2020_DHCa.P7. For NHOPIac measures, we use 2000_SF2.NPCT001A, 2010_SF1b.PCT10, and 2020_DHCa.P7.
Population density: We divide total population (see above) by land area (see below).
Median age: We use decennial tables 2000_SF1a.NP013A, 2010_SF1a.P13, and 2020_DHCa.P13. Median age is not available for 1990. For ACS data, we use estimate B01002_001.
Population shares by age category: Using the total population value above as the denominator, we use decennial tables 1990_STF1.NP11, 2000_SF1a.NP012B, 2010_SF1a.P12, and 2020_DHCa.P12.
About the Infrastructure data
Land and water areas: We convert shapefile land area and water area fields (see “Geographies” above) from square meters to square miles. The exception to this is the 1990 shapefiles, on which pre-calculated areas were missing, and we instead calculate areas programmatically.
Housing units: We use decennial tables 1990_STF1.NH1, 2000_SF1a.NH001A, 2010_SF1a.H1, and 2020_DHCa.H1.
Share of housing vacant (not occupied): We use decennial tables 1990_STF1.NH2, 2000_SF1a.NH003A, 2010_SF1a.H3, and 2020_DHCa.H3. For ACS data, we use group B25002.
Median home value: We use decennial tables 1990_STF1.NH23B and 2000_SF3a.NH076A. For ACS data, we use estimate B25077_001. In aggregations, we use each geography’s total occupied housing units for weighting.
Median gross rent: We use decennial tables 1990_STF3.NH43A and 2000_SF3a.NH063A. For ACS data, we use estimate B25064_001. In aggregations, we use each geography’s total occupied housing units for weighting.
Share of households cost-burdened: We use decennial tables 1990_STF3.NH51, 1990_STF3.NH58, 2000_SF3a.NH069A, and 2000_SF3a.NH094A. We follow the Census and U.S. Department of Housing and Urban Development convention of defining cost-burdened as housing cost representing 30 percent or more of household income. Households reporting no income or no cash rent are excluded from the denominator.
Share of housing owner-occupied: For the whole-population measure, we use decennial tables 1990_STF1.NH3, 2000_SF1a.NH003A, 2010_SF1a.H4, and 2020_DHCa.H4. For AIANa and NHOPIa measures, we use decennial tables 1990_STF1.NH9 (AIANa only), 2000_SF1a.NH014A, 2010_SF1a.H14, and 2020_DHCa.H4C. For AIANac and NHOPIac measures, we use 2000_SF4a.NHCT002A and 2010_SF1a.HCT2. AIANac and NHOPIac values for 2020 are currently unavailable.
Share of housing crowded: We use decennial tables 1990_STF1.NH22 and 2000_SF3a.NH020A. We follow the Census convention of defining crowded as more than one occupant per room, though we note the role of preference in housing arrangements and use the term neutrally.
Share of housing with complete kitchen facilities: We use decennial tables 1990_STF3.NH42 and 2000_SF3a.NH051A. For ACS data, we use group B25052. Only units that are occupied (that is, not vacant) are included.
Share of housing with complete plumbing facilities: We use decennial tables 1990_STF3.NH64 and 2000_SF3a.NH022A (2000). For ACS data, we use group B25048. Only units that are occupied (that is, not vacant) are included.
Household computer and Internet characteristics: Data collection related to these concepts began with the 2013 ACS. No decennial data have been collected.
About the Education data
K-12 enrollment: For the whole-population measure, we use decennial tables 1990_STF3.NP54 and 2000_SF3a.NP036D. For the AIANa-focused measure (in decennial data, only available in 1990), we use 1990_STF3.NP55 and ACS group B14007C.
Educational attainment: For the whole-population measure, we use decennial tables 1990_STF3.NP57 and 2000_SF3a.NP037C. For Native-focused measures, we use tables 1990_STF3.NP58 (AIANa only) and 2000_SF3a.NP148C. High school graduate values include equivalency credentials.
About the Economy data
Total civilian labor force: The total civilian labor force is the sum of employed and unemployed workers ages 16 and older. We use decennial tables 1990_STF3.NP70 and 2000_SF3a.NP150E.
Labor force participation: Labor force participation is the share of the population that is in the labor force (that is, the labor force divided by the population). For the whole-population measure, we use decennial tables 1990_STF3.NP70 and 2000_SFa.NP150C. For Native-focused measures, we use tables 1990_STF3.NP71 (AIANa only) and 2000_SF3a.NP150C (2000).
Employment-to-population ratio (EPOP; employment) and unemployment: The employment-to-population ratio is the number of employed workers divided by the population. Unemployment is the number of nonemployed people who are available and looking for work—or who expect recall to a previous job—divided by the number of people in the labor force. For these measures, we use the tables described above.
Median household income: For the whole-population measure, we use decennial tables 1990_STF3.NP80A and 2000_SF3a.NP053A. For Native-focused measures, we use tables 1990_STF4b.NPB65A (AIANa only) and 2000_SF3a.NP152A.
Per-person income: For the whole-population measure, we use decennial tables 1990_STF3.NP114A and 2000_SF3a.NP082A. For Native-focused measures, we use tables 1990_STF3.NP115A (AIANa only) and 2000_SF3a.NP157A.
Population experiencing poverty: For the whole-population measure, we use decennial tables 1990_STF3.NP117 and 2000_SF3a.NP087B. For Native-focused measures we use tables 1990_STF3.NP119 (AIANa only) and 2000_SF3a.NP159B.
Youth experiencing poverty: For the whole-population measure, we use decennial tables 1990_STF3.NP117 and 2000_SF3a.NP087C. For Native-focused measures, we use tables 1990_STF3.NP119 (AIANa only) and 2000_SF3a.NP159C.
Aggregate income: This measure describes total income for individuals ages 15 and older. For the whole-population measure, we use decennial tables 1990_STF3.NP116 and 2000_SF3a.NP083A and ACS group B19313. For Native-focused measures, we use decennial tables 1990_STF3.NP115 (AIANa only) and 2000_SF3a.NP158A, and groups B19313C (AIANa) and B19313E (NHOPIa).