Housing Crisis Rural Areas

County-level firm birth data are sourced from the U.S. Census Bureau's Business Dynamics (BDS) Statistics (Available at https://www.census.gov/data/datasets/time-series/econ/bds/bds-datasets.html), while labor force data by county are from the Bureau of Labor Statistics (BLS) (Available at https://www.bls.gov/lau/tables.htm#cntyaa). Both series span many decades, so the limiting series is the 5-year American Community Survey (ACS) (Available at https://www.census.gov/programs-surveys/acs.html or, more conveniently, at https://data.census.gov/), which has provided data on the number of housing units in a consistent manner only since 2010. To extend the time frame, note that the 2011 ACS estimates the number of housing units built between 2000 and 2004 (Table DP04 variable 0018E, labeled “Estimate!!YEAR STRUCTURE BUILT!!Built 2000 to 2004”).Additionally, the 2011 ACS also provides an estimate of housing units built in 2005 or later, whereas the 2014 ACS reports the total number of housing units built since 2010 (Table DP04 variable 0017E, labeled Estimate!!YEAR STRUCTURE BUILT!!Built 2005 or later” in the 2011 ACS and “Estimate!!YEAR STRUCTURE BUILT!!Total housing units!!Built 2010 or later” in the 2014 ACS). The difference between those was used to estimate the number of housing units built between 2005 and 2009. The added housing units between 2010 and 2014 were calculated by subtracting the total housing units in 2010 (2010 ACS) from those in 2014 (2014 ACS). The added housing units between 2015 and 2019 were obtained similarly (Table DP04 variable 0001E labeled “Estimate!!HOUSING OCCUPANCY!!Total housing units”).

The added labor force was also estimated by the difference between the total labor force in the respective time periods using BLS data. This paper’s focus is on business recruitment efforts. Therefore, firm entry was determined by summing the number of firm births over the years within each period. However, while the dataset reports counties with zero firm births, if the count is between 1 and 3, it is redacted to protect the firms’ confidentiality. The issue of missing values would be exacerbated since the values are combined in 5-year blocks. Thus, the missing values were replaced with a count of 1 firm birth. Robustness checks were conducted by substituting the missing values with zero births, and the results remained qualitatively unchanged.