Introduction:

The U.S. Census Bureau has been releasing county-to-county and county/minor civil division (MCD)-to-county/MCD migration flow estimates based on the American Community Survey (ACS) since 2012. [1] The first table package used data collected from 2005 through 2009 that composed the first ACS 5-year dataset.[2] For sequential 5-year ACS datasets, tables were crossed by selected characteristics. The 2008-2012 release was the first time flows from Puerto Rico municipios to U.S. counties were included. Flows between metropolitan statistical areas (MSAs) were introduced with the 2009-2013 dataset.[3] In addition, the county-to-county flows going back to 2006-2010 dataset are available through the Census Flows Mapper.[4] As of 2016, all ACS migration flows data for counties, MCDs, and MSAs were made available in the Census application programming interface (API) through the Census Developers site.[5]

*Data at the MCD level are only available for the strong-MCD states of Connecticut, Maine, Massachusetts, Michigan, Minnesota, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, Vermont, and Wisconsin.

The 2011-2015 Release:

The current release uses the first non-overlapping 5-year dataset since ACS migration flow tables crossed by characteristics were released. Therefore, the same demographic characteristics used for the 2006-2010 dataset are repeated for the current release: age, sex, race, or Hispanic origin.

The universe for all the characteristics is the population 1 year and over. Age is broken down into the following fifteen categories: 1 to 4 years, 5 to 17 years, 18 to 19 years, 20 to 24 years, 25 to 29 years, 30 to 34 years, 35 to 39 years, 40 to 44 years, 45 to 49 years, 50 to 54 years, 55 to 59 years, 60 to 64 years, 65 to 69 years, 70 to 74 years, and 75 years and over.

The four race categories consist of white alone, black or African American alone, Asian alone, and other race alone or two or more races. The three Hispanic origin categories are white alone, not Hispanic or Latino; not white alone, not Hispanic or Latino; and Hispanic or Latino.

Disclosure Avoidance:

Due to the amount of data released through standard 5-year ACS data products as well as the public use microdata samples (PUMS), and because sequential annual 5-year estimates contain overlapping samples; the Census Bureau must be vigilant in order to protect confidentiality of the respondents. Any data products that the Census Bureau releases must be reviewed and approved by the Disclosure Review Board based on disclosure avoidance rules to assure confidentiality of the respondents is protected according to Section 9 of Title 13 of the United States Code. It states “Neither the Secretary, nor any other officer or employee of the Department of Commerce or bureau or agency thereof, or local government census liaison may…make any publication whereby the data furnished by any particular establishment or individual under this title can be identified…”[6]

Record swapping, done to protect confidentiality, is applied to the ACS dataset used to create the flow files.[7] The additional disclosure avoidance measure of suppressing flows with small numbers of cases is applied for any flows with characteristics.[8] Flow counts and mover counts without characteristics, however, are not suppressed. The exact disclosure measures are determined by the Disclosure Review Board before each future product is released.[9]

Complementary suppression is applied to the MSA flows by characteristics so that suppressed county flows cannot be derived. If one and only one county-to-county flow of all the flows that aggregate up to a MSA-to-MSA flow is suppressed, then the MSA flow is suppressed. This can include small flows along with larger flows. If a MSA-to-MSA flow is suppressed for a particular characteristic, some information about the flow can be obtained from the MSA-to-MSA flow table without characteristics along with the county-to-county flows by that characteristic within each MSA that are not suppressed.  In addition to the MSA flows, complementary suppression is applied to county/MCD-to-county/MCD flows.

Since some flows are suppressed, summary statistics for each county, MCD, or MSA (e.g., number of nonmovers, movers to a different state) are included in the flow files. Tables crossed by selected characteristics are published annually as part of the ACS standard data products available through American FactFinder (AFF) for both current residence and residence 1 year ago geography. The selected characteristics do not cover the same characteristics used in this release. [10] The Census Bureau applies separate disclosure avoidance rules to these tables containing residence 1 year ago estimates. The universe for each table must have at least 50 unweighted cases for the Census Bureau to publish the table. The same rules are applied to the county-to-county, county/MCD-to-county/MCD, and MSA-to-MSA files. If there is not a sufficient number of cases, the summary statistics for county, MCD, or MSA of residence 1 year ago by characteristics are not shown.

Blank Values:

There are blank cells within the tables. An estimate may be blank for the following reasons.

1. The estimate is out of scope for the ACS sample – such as supplemental statistics for world regions.
2. The estimate does not exist – such as movers to a different county in the same state for the District of Columbia.
3. The supplemental statistics for county of residence 1 year ago or MCD of residence 1 year ago were suppressed due to disclosure avoidance (characteristic files only).

Besides blank estimates, flows with no records in the database and suppressed flows crossed by characteristics are not included in the files. The flows with no records are considered zero count estimates. The margins of error (MOE) for these zero count estimates cannot be calculated using the standard replicate variance formula.[11] For published ACS data products, the MOE for a zero count estimate is calculated using a different method. The standard errors for the zero count estimates are calculated by taking the square root of the product of the 90^th percentile k-value and the average weight for the state of residence. Estimates for previous residence use the U.S. average weight. The standard error is then multiplied by 1.645 in order to get the 90 percent MOE.[12] Before the 2012 ACS, only a single k-value was used. The methodology was changed to account for more nuanced k-values. This resulted in 6 different k-values for 5-year ACS. They are assigned based upon the total population of an area. For certain geographies, the total population is the official population estimate. If there is no official population estimate available, then the total population may be found in table B01003 through American FactFinder.

The average weights are calculated for each state of current residence and the k-value for the county of current residence is used. All estimates for residence 1 year ago and current residence estimates for metro areas that cross state lines should use the average weight for the United States.

Coverage and Group Quarters Population:

The American Community Survey covers the entire population residing in the United States and Puerto Rico, both in housing units and group quarters facilities. Each year, independent housing unit address samples are selected for each county equivalent in the United States and Puerto Rico. Samples of group quarters facilities and persons in group quarters are done at the state level, including Puerto Rico and the District of Columbia. In order to provide a better representation of group quarters estimates over smaller areas, whole person imputation into not-in-sample group quarters facility was started with the 2011 ACS, including the 2007-2011 ACS 5-year estimates.[13] To better preserve the relationship between current and previous residence for these imputed cases, a recipient’s county of residence 1 year ago is changed to the current county of residence if the donor moved within the same county; otherwise, the donor’s county of residence 1 year ago is used for the imputation.

The group quarters population includes people living in correctional institutions, juvenile detention facilities, nursing homes, other long-term care facilities, college dormitories, military facilities, and other noninstitutionalized facilities. An occasional anomaly in the number of movers between a county pair can occur due to whole person group quarters imputation and higher rates of inter-county migration for people in group quarters compared to the general public. The 2011-2015 ACS inter-county mover rates for group quarters are shown in the table below. The estimated group quarters population for a county can be found in detailed table B26001 on American FactFinder.

Source: U.S. Census Bureau, 2011-2015 American Community Survey, Subject Table S2601B

Application Programming Interface (API):

In 2016, all ACS migration flows data for counties, MCDs, and MSAs were made available in the Census API through the Census Developers site.[14] The API files contain the same estimates that were previously released in table form and Census Flows Mapper, but the data are provided in a different layout to allow developers to design web and mobile apps that include Census Bureau statistics. See the API website for more information.

Format of the Files:

There are separate text and Excel files for flows between counties, flows between metropolitan statistical areas (MSAs), and flows between minor civil divisions (MCDs) in some selected states. In several states, minor civil divisions are county subdivisions that are the primary governmental or administrative divisions of a county. In 12 states, the MCD governments serve as general-purpose local governments similar to incorporated municipalities.[15] In some of these states, county governments are limited or non-existent; therefore, MCDs are used as the main substate aggregate unit rather than counties. These 12 states are assigned MCD codes for residence 1 year ago during the geocoding process for ACS.

County-to-County Flow Files with No Characteristics

The county-to-county flow files are provided in two formats. There is one fixed field length national text file sorted by current residence geography. There is also an Excel file (.xlsx) containing a worksheet for each state of current residence, including Puerto Rico and the District of Columbia, and another Excel file containing a worksheet for each state of residence 1 year ago.

Besides the county of current residence, county of residence 1 year ago, and the number of movers between the two, the files also contain additional geographical mobility estimates for each county, along with the margin of error (MOE) at the 90 percent confidence level. (For further information about the geographies used for the files see Appendix A.)

The layout of the text file is as follows:

County-to-County Net and Gross Migration Files (No Characteristics)

There are two files, text and Excel, that contains flow, counterflow, net migration, and gross migration for all county pairs (Geography A and Geography B). The Excel file has a worksheet for each state. Both files are sorted by Geography A. The flow shows the inmigration from Geography B to Geography A. The counterflow shows the outmigration from Geography A to Geography B. The net migration is the flow minus the counterflow, and the gross migration is the flow plus the counterflow.

The layout for the text file is as follows:

County-to-County Flow Files by Characteristics

There is a set of county-to-county files, text and Excel, for each of the three characteristics. The county-to-county characteristic flow files have the same fields as the county-to-county flow files with the additional field for characteristic code in positions 14-15. The field contains relationship to householder code, household type code, or housing tenure code depending upon the file.

The layout of the text file is as follows:

County/MCD-to-County/MCD Flow Files

The County/MCD-to-County/MCD files are similar to the county-to-county files except that MCDs are used instead of counties for Connecticut, Maine, Massachusetts, Michigan, Minnesota, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, Vermont, and Wisconsin.

There is one fixed field length national text file sorted by current residence geography. There is also an Excel file (.xlsx) containing a worksheet for each state of current residence, including Puerto Rico and the District of Columbia, and an another Excel file containing a worksheet for each state of residence 1 year ago.

In addition to the county or MCD of current residence, county or MCD of residence 1 year ago, and the number of movers between the two, the files also contain additional geographical mobility estimates for each county or MCD, along with the margin of error (MOE) at the 90 percent confidence level. (For further information about the geographies used for the files see Appendix A.)

The layout of the text file is as follows:

County/MCD-to-County/MCD Net and Gross Migration Files (No Characteristics)

There are two files, text and Excel, that contains flow, counterflow, net migration, and gross migration for all county/MCD pairs (Geography A and Geography B). The Excel file has a worksheet for each state. Both files are sorted by Geography A. The flow shows the inmigration from Geography B to Geography A. The counterflow shows the outmigration from Geography A to Geography B. The net migration is the flow minus the counterflow, and the gross migration is the flow plus the counterflow.

The layout for the text file is as follows:

County/MCD-to-County/MCD Flow Files by Characteristics

There is a set of county/MCD-to-county/MCD files, text and Excel, for each of the three characteristics. The county/MCD-to-county/MCD characteristic flow files have the same fields as the county/MCD-to-county/MCD flow files with the additional field for characteristic code in positions 24-25. The field contains relationship to householder code, household type code, or housing tenure code depending upon the file.

The layout of the text file is as follows:

MSA-to-MSA Flow Files with No Characteristics

The MSA-to-MSA flow files are provided in two formats. There is one fixed field length national text file sorted by current residence geography. There is also an Excel file (.xlsx) containing flows between all the MSAs in the United States and Puerto Rico and the District of Columbia.

Besides the MSA of current residence, MSA of residence 1 year ago, and the number of movers between the two, the files also contain additional geographical mobility estimates for each MSA along with the margin of error (MOE) at the 90 percent confidence level. (For further information about the geographies used for the files see Appendix A.)

The layout of the text file is as follows:

MSA-to-MSA Net and Gross Migration Files (No Characteristics)

There are two files, text and Excel, that contains flow, counterflow, net migration, and gross migration for all msa pairs (Geography A and Geography B). The Excel file has a single worksheet for all flows. Both files are sorted by Geography A. The flow shows the inmigration from Geography B to Geography A. The counterflow shows the outmigration from Geography A to Geography B. The net migration is the flow minus the counterflow, and the gross migration is the flow plus the counterflow.

The layout for the text file is as follows:

MSA-to-MSA Flow Files by Characteristics

There is a set of MSA-to-MSA files, text and Excel, for each of the three characteristics. The MSA-to-MSA characteristic flow files have the same fields as the MSA-to-MSA flow files with the additional field for characteristic code in positions 12-13. The field contains relationship to householder code, household type code, or housing tenure code depending upon the file.

The layout of the text file is as follows:

Appendix A: Geography

County Equivalents

Counties are the primary legal division in most states, but a few states have equivalent divisions known by different names. In order to get a complete partition of the United States and Puerto Rico, the following divisions are treated as county equivalents in the files.

Borough (Alaska)

Census Area (Alaska)

City and Borough (Alaska)

County (All except Alaska, the District of Columbia, Louisiana, and Puerto Rico)

The District of Columbia

Independent City (Maryland, Missouri, Nevada, Virginia)[16]

Municipality (Alaska)                                                                 

Municipio (Puerto Rico)

Parish (Louisiana)

Minor Civil Divisions

The county/MCD-to-county/MCD files include minor civil division estimates for the 12 states in which they also serve as general-purpose local governments. The 12 states have various names for their minor civil divisions.

Connecticut (City, Town)

Maine (City, Gore, Indian Reservation, Plantation, Town, Unorganized Territory)

Massachusetts (City, Town)

Michigan (Charter Township, City, Township)

Minnesota (City, Township, Unorganized Territory)

New Hampshire (City, Grant, Location, Town, Township)

New Jersey (Borough, City, Purchase, Town, Township, Village)

New York (Borough, City, Indian Reservation, Town)

Pennsylvania (Borough, City, Town, Township)

Rhode Island (City, Town)

Vermont (City, Gore, Grant, Town)

Wisconsin (City, Town, Village)

Puerto Rico

Prior to the 2008 ACS, the stateside questionnaire only requested respondents to answer “Puerto Rico” if they lived in Puerto Rico 1 year ago. The Puerto Rico questionnaire asked for a municipio or a U.S. county for movers, along with other geographic information; therefore, municipio level migration flow data from Puerto Rico to the United States are not available prior to 2008 but are available for flow from the United States to each Puerto Rico municipio.

The question was revised in 2008 to ask for the same geographic information on the Puerto Rico and stateside questionnaire allowing estimates to be calculated for both flow directions. Because of this limitation, prior ACS county-to-county migration flow files aggregated flows from Puerto Rico to the United States to “Puerto Rico” rather than individual municipios. The 2008-2012 files are the first to include flows from Puerto Rico municipios to U.S. counties. Puerto Rico municipios are treated as equivalent to U.S. counties.

U.S. Island Areas and Foreign Countries

Outmigration from the United States and Puerto Rico to U.S. Island Areas or foreign countries is not available from the American Community Survey since only housing units and group quarters (e.g., college dormitories, military barracks, prisons) within the United States and Puerto Rico are sent questionnaires. The American Community Survey does collect data for U.S. Island Area or Foreign Country of residence 1 year ago. The files include inmigration from outside the United States and Puerto Rico aggregated to U.S. Island Areas and foreign region. A three letter code is used to identify these areas.

U.S. Island Areas (ISL):

American Samoa, Baker Island, Guam, Howland Island, Jarvis Island, Johnston Atoll, Kingman Reef, Northern Marianas Islands, Midway Islands, Navassa Island, Palmyra Atoll, U.S. Virgin Islands, Wake Island, U.S. Island Areas not specified.

Europe (EUR):

Albania, Andorra, Austria, Azores Islands, Belarus, Belgium, Bosnia and Herzegovina, Bulgaria, Croatia, Czechoslovakia, Czech Republic, Denmark, England, Estonia, Faroe Islands, Finland, France, Germany, Gibraltar, Greece, Guernsey, Hungary, Iceland, Ireland, Isle of Man, Italy, Jan Mayen, Jersey, Kosovo, Latvia, Liechtenstein, Lithuania, Luxembourg, Macedonia, Malta, Madeira Islands, Moldova, Monaco, Montenegro, Netherlands, Northern Ireland, Norway, Poland, Portugal, Romania, Russia, San Marino, Scotland, Serbia, Slovakia, Slovenia, Spain, Svalbard, Sweden, Switzerland, Ukraine, United Kingdom, USSR, Vatican City, Wales, Yugoslavia, Europe not specified.

Asia (ASI):

Afghanistan, Armenia, Azerbaijan, Bahrain, Bangladesh, Bhutan, Brunei, Cambodia, China, Cyprus, East Timor, Georgia, Hong Kong, India, Indonesia, Iran, Iraq, Israel, Japan, Jordan, Korea, Kazakhstan, Kyrgystan, Kuwait, Laos, Lebanon, Macau, Malaysia, Maldives, Mongolia, Myanmar (Burma), Nepal, North Korea, Oman, Pakistan, Paracel Islands, Philippines, Qatar, Saudi Arabia, Singapore, South Korea, Spratley Islands, Sri Lanka, Syria, Taiwan, Tajikistan, Thailand, Turkey, Turkmenistan, United Arab Emirates, Uzbekistan, Vietnam, Yemen, Asia not specified.

Northern America (NAM):

Bermuda, Canada, Greenland, St. Pierre & Miquelon, North American not specified

Central America (CAM):

Belize, Costa Rico, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Central America not specified

Caribbean (CAR):

Anguilla, Antigua and Barbuda, Aruba, Bahamas, Barbados, Bonaire, British Virgin Islands, Cayman Islands, Cuba, Curaçao, Dominica, Dominican Republic, Grenada, Guadeloupe, Haiti, Jamaica, Martinique, Montserrat, Netherland Antilles, Saba, St. Barthelemy, St. Kitts-Nevis, St. Lucia, St. Vincent and the Grenadines, Sint Eustatius, Sint Maarten, Trinidad and Tobago, Turks and Caicos Islands, West Indies not specified

South America (SAM):

Argentina, Bolivia, Brazil, Chile, Colombia, Ecuador, Falkland Islands, French Guiana, Guyana, Paraguay, Peru, Suriname, Uruguay, Venezuela, South American not specified

Africa (AFR):

Algeria, Angola, Benin, Botswana, British Indian Ocean Territory, Burkina Faso, Burundi, Cameroon, Cabo Verde, Central African Republic, Chad, Comoros, Congo, Democratic Republic of Congo, Djibouti, Egypt, Equatorial Guinea, Ethiopia, Eritrea, Europa Island, Gabon, Gambia, Ghana, Glorioso Islands, Guinea, Guinea-Bissau, Ivory Coast, Juan de Nova Island, Kenya, Lesotho, Liberia, Libya, Madagascar, Malawi, Mali, Mauritania, Mauritius, Mayotte, Morocco, Mozambique, Namibia, Niger, Nigeria, Reunion, Rwanda, Sao Tome and Principe, Senegal, Seychelles, Sierra Leone, Somalia, South Africa, South Sudan, St Helena, Sudan, Swaziland, Tanzania, Togo, Tromelin Island, Tunisia, Uganda, Western Sahara, Zambia, Zimbabwe, Africa not specified

Oceania and At Sea (OCE):

Australia, Christmas Island, Cook Islands, Coral Sea Islands, Heard and McDonald Islands, Fiji, French Polynesia, Kiribati, Marshall Islands, Micronesia, Nauru, New Caledonia, New Zealand, Niue, Norfolk Island, Palau, Papua New Guinea, Pitcairn Islands, Solomon Islands, Tokelau, Tonga, Tuvalu, Vanuatu, Wallis and Futuna Islands, Samoa, Oceania not specified, At sea

References:

Benetsky, Megan and Kin Koerber. 2005-2009 American Community Survey County-to-County Migration Files.

<https://www.census.gov/library/working-papers/2012/demo/benetsky-02.html>

Bortlein, Celia G. and Alan K. Peterson. Evaluation Report Covering Residence 1 Year Ago, 2007.

<https://www.census.gov/library/working-papers/2007/acs/2007_Boertlein_01.html>

U.S. Census Bureau. American Community Survey Design and Methodology (January 2014). Washington, D.C.: U.S Government Printing Office, 2014.

<http://census.gov/programs-surveys/acs/methodology/design-and-methodology.html>

U.S. Census Bureau. Puerto Rico Community Survey Multiyear Accuracy of the Data (5-year 2011-2015).

<https://www2.census.gov/programs-surveys/acs/tech_docs/accuracy/MultiyearACSAccuracyofData2015.pdf>

Zayatz, Laura. “Disclosure Avoidance Practices and Research at the U.S. Census Bureau: An Update.” Research Report Series 2005-06. Statistical Research Division, U.S. Census Bureau, Washington, D.C., August 31, 2005.

<http://census.gov/srd/papers/pdf/rrs2005-06.pdf>

[1] For additional background information about historical migration flows, see <https://www.census.gov/library/working-papers/2012/demo/benetsky-02.html>. For information about the various characteristics, see <http://www2.census.gov/programs-surveys/acs/tech_docs/subject_definitions/2015_ACSSubjectDefinitions.pdf>.

[2] The ACS program combines consecutive yearly datasets to increase the sample size and provide reliable estimates for smaller areas. Even though estimates are produced for 1-year datasets, they are limited to geographic areas of at least 65,000 population. Only the 5-year datasets provide estimates for all counties and county equivalents in the United States.

[3] The MSA delineations are from February 2013. See <https://www.census.gov/population/metro/> for more information.

[4] The Census Flows Mapper <http://flowsmapper.geo.census.gov/> is an interactive, online mapping application.

[5] The Census Developers site is located at <https://www.census.gov/developers/>.

[6] For more information on how Census data are protected see <http://census.gov/about/policies/privacy/data_protection.html>.

[7] Data swapping was the main procedure used for protecting Census 2000 tabulations and is also used for ACS tabulations. In each case, a small percentage of household records are swapped. Pairs of households in different geographic regions are swapped. The selection process for deciding which households should be swapped is highly targeted to affect the records with the most disclosure risk. Pairs of households that are swapped match on a minimal set of demographic variables. All data products are created from the swapped data files (American Community Survey Design and Methodology, 2015).

[8] County and MCD characteristic flows containing only one or two people from different households, only one or two people in group quarters, or one person in group quarters and the rest from a single household are suppressed. People are only counted if they are in universe (e.g., population 1 year and over in households for relationship to householder).

[9] Prior to the 2010-2014 release, some suppressed county and MCD flows were summed up to the state of residence 1 year ago. Those estimates are no longer included in the tables.

[10] The tables in AFF are crossed by age (B07001 and B07401), median age (B07002 and B07402), sex (B07003 and B07403), race and Hispanic origin (B07004A-I and B07404A-I), citizenship status (B07007 and B07407), marital status (B07008 and B07408), educational attainment (B07009 and B07409), individual income (B07010 and B07410), median individual income (B07011 and B07411), poverty status (B07012 and B07412), and housing tenure (B07013 and B07413). The stubs for the characteristic tables in AFF refer to movement between states and counties and not MCDs or MSAs.

[11] For more information concerning the calculation of margins of error, see chapter 12 of Design and Methodology: American Community Survey. <http://www2.census.gov/programs-surveys/acs/methodology/design_and_methodology/acs_design_methodology_report_2014.pdf>

[12] More on the model to compute ACS margins of error for zero-estimate counts can be found at <http://www2.census.gov/programs-surveys/acs/tech_docs/user_notes/KValueUserNote.pdf>.

[13] For more information on ACS group quarters small area estimation, see <http://www2.census.gov/programs-surveys/acs/tech_docs/user_notes/GQSAE_User_Note.pdf>.

[14] The Census Developers site is located at <https://www.census.gov/developers/>.

[15] For more information on geographic terms and concepts, see <http://census.gov/geo/www/reference.html>.

[16] In previous county-to-county flows products, some counties and independent cities were collapsed into single entities due to cognitive and geographic coding problems with migration data for places. (For the Census 2000 list of combined Virginia counties and independent cities see Appendix B in Migration DVD documentation at <http://census.gov/population/www/cen2000/migration/mig_dvd.html>).