Vacancy in ACS vs Decennial

They use a different methodology - the decennial doesn't have sampling error, but it is a point in time estimate. The ACS data is collected monthly as a 5-year average and can have wide margins for small areas.

Here's a working paper from a while ago that likely is still helpful:
www.census.gov/.../2011_Griffin_03.pdf

If you are a nonprofit or govt, you can also use the USPS data from HUD (tract level)
- www.huduser.gov/.../usps.html

Here's a background paper on that source: www.neighborhoodindicators.org/.../valassis-lists-data-indicator-population-recovery-new-orleans-area

Hi All, Vacancy data derived from USPS sources may not be reliable for determining vacancy. 

 

From the AAPOR Task Force Report on Address-based Sampling...

 

"The USPS maintains an address vacancy indicator. In 2010, there were about 13 million

addresses indicated as vacant in the AMS (in the raw DSF delivered to the Census Bureau).

However, recent studies have found between 38% and 41% of units classified as vacant to be

occupied (Amaya et al. 2014; Kalton et al. 2014). A housing unit that is vacant may quickly

become occupied. According to USPS guidelines, an address must be unoccupied for 90 days to

be classified as vacant (Iannacchione 2011). Given the difficulty in maintaining an accurate and

timely vacancy indicator and the coverage concerns with using this variable in frame creation,

caution should be exercised in removing addresses classified as vacant."

 

 

https://www.aapor.org/getattachment/Education-Resources/Reports/AAPOR_Report_1_7_16_CLEAN-COPY-FINAL-(2).pdf.aspx

Thanks Kathy and Joseph,
We're definitely aware of the merits of the USPS-sourced vacancy data (we source it through Valassis). These concerns over it are interesting, thanks for sending them along.

But in terms of Census vacancy data, I guess another way to phrase my question is, in a small area, given the choice of the most recent ACS data, and the last decennial data, are there circumstances where you would you consider forgoing the much more recent ACS data, and instead go with the older decennial data?

Thanks,
Bernie

I don't know you're exact use case but 9 years back is a long way, so I'd tend to go with ACS.

Another consideration is a factor not present in a significant way in 2010, which is the effect of AirBnB on vacancy rates. Communities with a large AirBnB presence are almost certainly going to have a higher vacancy rate now than in 2010, all other things factors being held equal. If this is the case on the community of interest then using a vacancy rate from 2010 to describe current conditions provides an incomplete picture of the situation. In the case of Cambridge, MA, where I work, I believe that AirBnB might be raising the rental vacancy rate by as much as 1-2%, which is huge in our market.

Another consideration is a factor not present in a significant way in 2010, which is the effect of AirBnB on vacancy rates. Communities with a large AirBnB presence are almost certainly going to have a higher vacancy rate now than in 2010, all other things factors being held equal. If this is the case on the community of interest then using a vacancy rate from 2010 to describe current conditions provides an incomplete picture of the situation. In the case of Cambridge, MA, where I work, I believe that AirBnB might be raising the rental vacancy rate by as much as 1-2%, which is huge in our market.

Thanks, Kathy, that's helpful. And Cliff, that is fascinating. According to ACS, vacancy increased by 67.03% in Cambridge from 2008-2012 to 2013-2017. I'm still concerned about the margins of error with that data, though.

Another factor in Cambridge is a lot of new apartment construction. This could affect vacancy in 2 ways.

It is still unclear to me at what point the ACS begins to count a unit under construction as vacant. One definition I believe I have seen is when the unit is "tight to the weather", ie doors, widows and roof are in place. If this is accurate then we are going to have a lot of vacant units that aren't really habitable yet.

The other factor is that when you drop 200-500 brand new units into a market at once then absorption takes time. If you multiply this by 3 or 4 buildings at a time on a rolling basis then you will probably see a boost in vacancy due to market friction, not a decline in need or desirability.