Improving Internal Migration Data for State and County Population Estimates: Can the American Community Survey Inform?

Hyo C. Park, U.S. Census Bureau
Bashiruddin Ahmed, U.S. Census Bureau

Internal migration is one of the major components of population change at both the state and county levels. Currently, estimates of state-to-state and county-to-county migration are produced using the annual IRS tax return files. The IRS-based estimates have several disadvantages: 1) the IRS data represent about 87 percent of the population, not the entire population universe; 2) the estimates are produced based on migration counts of filers and their dependents, assuming that the dependents and non-filers have the same migration rate as filers; and 3) the race and Hispanic origin information of dependents is assumed to be the same as the filers. To improve data on internal migration, the Census Bureau staff is conducting research on integration of migration information from Census 2000, administrative records, and survey results, particularly the American Community Survey. As a first step towards integration, this paper identifies differences in internal migration developed from the 2000 IRS and the Census 2000 Supplementary Survey (C2SS) data. Specifically, the paper addresses the following questions: 1) how different is the IRS-based migration from the C2SS-based migration? 2) which characteristics show the largest migration differences, and at what geographic levels? 3) are the migration differences mainly due to sampling errors? and 4) what proportion of outmigrants from the IRS file moved to the C2SS sampled counties, particularly the self-representing counties? The paper concludes with a discussion of a general model of estimating gross out- and in-migration for filers and non-filers together, using the ACS and IRS data. This will help us prepare a best set of migration estimates by characteristics that can be used in the cohort component technique for producing state and county population estimates.