U.S. Census Bureau

 Small Area Income & Poverty Estimates

 Model-based Estimates for States, Counties, & School Districts

Model Input Data


Several data sources are used in producing the SAIPE program estimates, though not all of these data are publicly available. Specific details can be found in Information about Data Inputs. The links below provide the aggregate tax and food stamp data used in our models.

Downloadable data files:

IRS data

The aggregate income data available on this site have been approved for public release by the Internal Revenue Service, Statistics of Income Division (IRS/SOI). Tax return data on individuals is protected by Title 26 U.S.C. and is never published. The data include:

Tax files for 1989 through 1998 contain data derived from returns filed through the first 9 months of the year following that tax year; the tax files for subsequent years contain the returns for all 12 months. Data users need to be aware that these data may not match data from other sources due to differences in definitions, and that the SAIPE program only uses tax returns for the relevant income year (IRS/SOI include all tax returns filed in a calendar year in their tabulations, regardless of the reference income year). Also note that starting in 1995, total exemptions include children living away from home (prior to 1995, only children at home were counted), exclude parent exemptions, and exclude "other" exemptions. This changes all variables that employ the number of exemptions in their calculation. In addition, tax returns claiming no exemptions -- i.e., those who are claimed as dependents on other returns -- are excluded from the calculation of all variables (except median income). For more information on how the SAIPE program uses the IRS data, please go to federal income tax returns.

Food Stamp data

The food stamp data represent the number of participants in the Food Stamp Program for each state, county, and the District of Columbia from 1981 to 2004. The state level files contain the number of food stamp recipients by month, and the county level files contain the number of recipients in July of each year.

Outliers of the food stamp data at the state level are smoothed on the basis of time series analysis. Outliers are smoothed starting in January of 1981 except for NE, NJ, NM, RI and TN, where smoothing starts in January of 1992; FL and WV, where smoothing starts in January of 1993; WA, where smoothing starts in January of 1995; and WY, where smoothing starts in March of 1997.

The number of food stamp recipients in Alaska and Hawaii are adjusted downward because the income eligibility thresholds for these states are higher than they are for states in the continental U.S.

The county food stamp totals are raked to the state totals which are also provided in the data files. These values may not match those in the state files (except for the most recent years) because these data are used in the actual production for the specified year. Also note, for the 1995 poverty estimates and onward, the state totals in the county food stamp files are based on the average number of food stamp recipients over twelve consecutive months. For example, the 2000 model uses an average over the period July 2000 through June 2001.

More information on how the SAIPE program uses food stamp data in the models can be found at food stamp recipients.

Poverty Universe data

The poverty universe is made up of persons for whom the Census Bureau can determine poverty status (either "in poverty" or "not in poverty"). The poverty universe used for the 2005 SAIPE estimates conceptually matches the poverty universe of the 2005 American Community Survey (ACS). The poverty universe estimates are not the same as the intercensal population estimates from the Census Bureau’s Population Estimates Program. They are derived estimates that differ from intercensal population estimates by the following points:

Since the Census Bureau surveys typically ask income questions to persons age 15 or older, if a child under age 15 is not related by birth, marriage, or adoption to a reference person within the household (for example, foster children), we do not know the child's income and therefore cannot determine his or her poverty status. Persons living in group quarters housing are not included in the 2005 poverty universe since the 2005 ACS did not survey from group quarters populations.

The poverty universes provided on this web site are the estimates used in the production of the SAIPE program's poverty rates for the given year and are not updated to reflect any successful challenges or other subsequent updates (annual or otherwise) to the published population estimates from the Population Estimates Program. Due to their method of construction, the poverty universe estimates are sometimes larger than the official population estimates. This happens partially because the age groups for the SAIPE program's poverty universe estimates do not in all cases match the age groups used as population controls in estimating the ACS.

Prior to the estimates for 2005, the poverty universe data was derived from the Annual Social and Economic Supplements of the Current Population Survey. Please see the denominators for model-based poverty rates webpage for details on how SAIPE poverty universes are produced.

Source: U.S. Census Bureau, Data Integration Division, Small Area Estimates Branch
For assistance, please contact the Demographic Call Center Staff at 301-763-2422 or 1-866-758-1060 (toll free) or visit ask.census.gov for further information.