Creating a "new rents" variable


I am wondering if anyone has been able to generate a variable using in-migration to a county and rent asked or paid, such that the variable created is a measure of what new rents in a county are? I am trying to figure out how rents change after a disaster, so determining the new asking prices is essential. If it is even possible, would the samples be too small to report?


This is technically possible using the variables MIGRATE1, STATEFIP, COUNTYFIP, MIGPLAC1, MIGCOUNTY1, and RENT. There are a number of limitations you should be aware of, though:

  1. About 40% of respondents are in geographic areas that cannot be identified to the county level, because the county has too small of population. These are identified by COUNTYFIP==0 or MIGCOUNTY1==0. For the current residence, the lowest geography is PUMA, which often comprise multiple counties. For last year, the lowest geography is MIGPUMA1, which is a coarser geography.

  2. The sample sizes per county (for identifiable counties) range from 769 to 101,631 in the 2017 ACS. You could combine multiple years to increase the sample, or use a 5-year ACS sample.

  3. If you look only at individuals in identifiable counties who moved from other identifiable counties, the sample sizes range from 33 up to 7360.

  4. About 25% of respondents in the sample paid any rent.

Another thing to consider is to use just the MIGRATE1 variable. This indicates whether a respondent resides in the same house as one year previously, or not. This would eliminate the need to consider last year’s county.