Scan/US, Inc. uses a number of government data series released at the US, State and County level to guide the demographic estimates and projections for all US counties. The Census Bureau publishes the Current Population Survey (CPS) annually; the most current data point in the CPS is 12 to 18 months prior to the publication date of the CPS. These series include basic counts: population, housing units and group quarters population, as well as detailed cross‐tabulations of population by age, sex, race and Hispanic origin. The Census Bureau's National Population Projection Series, currently 2012-2060, provides the detailed national annual projection of population by age, sex and race framework on which the lower level geography estimates are hung. The Bureau of Labor Statistics (BLS) provides the annual county‐level Labor Force Status series as well as current quarterly unemployment reports. The Bureau of Economic Analysis produces a county series on personal income.
In addition to these important traditional control series, the Census Bureau implemented the American Community Survey (ACS) program. The ACS program has replaced the Census Long Form, which was distributed to a 7% sample of US households as a part of every census until the 2010 Census. The ACS five‐year pooled survey now provides a reliable detailed socioeconomic profile for states, counties, county subdivisions, places, census tracts and block groups. The 2012 ACS 5 year pooled survey, published in late 2013, paired with the 2010 100% Census data for the same Census geography forms a baseline on which the annual demographic estimates and projections are based.
The US Postal Service releases monthly delivery counts for all postal carrier routes in the United States. These residential and business delivery counts provide a current measure of households and businesses receiving mail across all 242,000 carrier routes in the United States. Carrier Route delivery statistics, summarized to the county level, provide a household data point for the estimate year. Thus Scan/US can identify counties experiencing rapid change which are not reflected in the CPS series due to its 12 to 18 month lag behind its publication date.
The Census Bureau publishes an annual series of detailed household income tables for the United States from which Scan/US, Inc. generates an income category shift model for moving ACS five year income distributions forward over time. The ACS 5 year pooled income profiles at the lower levels of census geography are used to inform the estimates and projections of structural changes that cannot be captured in the US level income shift model.
Getting it right at the county level is only part of the battle. The real challenge is to map that change onto the landscape at a geographic resolution that supports location‐based decisions down to local neighborhoods. This requires being able to track household change down to the level of a few city blocks—city blocks that were not even defined in Census 2010. Scan/US, Inc. over the last two decades has developed and refined a proprietary cartography called Scan/US MicroGrids that reflect change down to 1/16th of a square mile. The MicroGrid cartography is a network of cells of three sizes: large ≈ 1 square mile, medium ≈ ¼ sq. mile, and small ≈ 1/16th sq. mile. MicroGrids exist where there is either housing or business development on the landscape, and grid size is determined by the density of that development. The smaller the grid, the denser the development contained by the grid.
The MicroGrid cartography is updated annually. The key ingredients in this process are a current national mailing list of deliverable residential and business addresses summarized to ZIP+4s, and latitude/longitude coordinates for each ZIP+4. The 40 million or so ZIP+4s, with their current residential and business deliveries, are poured into the existing MicroGrid network, resulting in the addition of grids in new developing areas, or the subdivision of larger grids to better define the housing distribution in growing neighborhoods. The end results are a MicroGrid cartography that represents the current distribution of households and businesses throughout the United States down to 1/16th of a square mile.
The current ZIP+4 residential counts are summarized to block groups within counties, and are used to calculate the participation rate of the block groups in the distribution of the county demographic estimates and projections. The base demographic profiles of the block groups are normalized to their new delivery count population and household estimates, and then the block groups within the county are balanced across all demographic dimensions to the county estimates. The block group’s detailed demographic profile then gets distributed to their blocks by normalizing the blocks’ base demographics to their new household and population controls and balancing each demographic dimension to the block group controls. The resulting block estimates are then used to build the demographic profiles of the MicroGrids through a block‐to‐MicroGrid cross reference which has household weights based on the ZIP+4 assignments.
All of this works fine if you are dealing with a block group that has a solid base demographic profile to build on. However, when dealing with block groups which had few or no households in the base year, but which are estimated to have gained a significant number of households over the ensuing years, some creativity is required. Under these circumstances, ScanUS recruits the most likely demographic profile from an adjacent block group in order to imbue the new households with the demographic profile of the neighborhood.