How crime affects the migration decisions of individuals is important for two main reasons. First, increases in crime may cause individuals to leave a locality, which erodes the tax base required to address further increases in criminal justice and public safety initiatives. Additionally, crime may inflict other externalities, since long-distance moves are costly.
While a few previous papers (Cullen and Levitt, 1999; Ellen and O’Regan, 2010) have estimated the effect of crime on net population changes, my paper more explicitly looks at the distance of migration undertaken in response to increases in crime. In order to measure the explicit costs of crime due to relocations, I extend these previous papers by decomposing the net migration of a county into gross migration flow rates.
Because I focus on migration within and outside a metropolitan statistical area (MSA), which is an area with a high population density and close economic ties, I measure four different gross migration flows: migration out of the county, to other counties within the MSA; migration out of the county, to counties outside the MSA; migration into the county, from other counties within the MSA; and migration into the county, from other counties outside the MSA. Using these four variables as outcomes, I test to what extent crime causes the gross migration flows to change, using a fixed-effects research design and including county-specific trends.
In order to measure these migration flows, I use the county-to-county migration flows from the Internal Revenue Service (IRS) Statistics of Income (SOI) files from 1984-2010. These files report migration of individuals between counties, and they allow me to count how many individuals moved within a metropolitan area and outside a metropolitan area.
I find that increases in crime tend to cause individuals to leave the metropolitan area entirely, rather than to relocate to a different county in the same MSA. However, my results are much smaller than previous estimates, which suggests that the costs of crime are overstated by these papers.
In addition to measuring how crime impacts relocation decisions, I also measure how crime changes the composition of the population in a county. Using the SEER population database, I measure the one-year percentage change in the population size for a number of subgroups: ages 0-18, 19-44, 45-59, 60 and older, as well as white and black individuals. These one-year percentage changes in the population size can be thought of as proxies for the subgroup-specific migration rate, and measure how responsive these specific groups are to changes in the crime rate.
Using these measures, I find significant differences in responsiveness across demographic groups. First, I found that individuals ages 19-59 move out of areas more quickly in response to crime, while older individuals are not responsive. More importantly, I find that white individuals respond to crime rates in a statistically significant way; according to my coefficient estimates, whites are about nine times more likely to migrate as a response to an increase in crime rates as blacks.
Overall, my results suggest that crime is costly to society because individuals relocate when they otherwise would not have done so, but that it is less costly in terms of migration costs than previous papers would have suggested. However, crime also has important consequences on the population composition of areas, with white individuals moving out disproportionately more in response to crime shocks, while black individuals seem to stay in place. These findings corroborate Patrick Sharkey’s recent work on the relative immobility of black individuals, which may have long-term impacts in terms of economic opportunity and quality of life.