Second homes estimate

Hi,

This article: The Nation’s Stock of Second Homes | Eye On Housing provides estimates for the number of second homes. They mention that they use ACS data “In terms of methodology, this analysis focuses on the number and location of second homes that would be qualified for the home mortgage interest deduction by individuals using the Census Bureau’s 2020 American Community Survey (ACS). It does not account for homes held primarily for investment or business purposes.”

I was wondering what data is that? There is a second home variable in IPUMS but it only has data in 1970. There is also a second mortgage variable, but that seems to just capture a second mortgage on the same house (so not a measure of a second home).

Please let me know if anyone has any idea what type of data they are using in that article.

Thanks a lot in advance.

Best,
Ricardo

Dear Ricardo,

Based on my understanding of blog post you link to, the NAHB probably started with table B25004. Vacancy Status from the 2016-2020 ACS 5-year Summary file. The map assigns all counties to a bin, and the only way to identify all counties in the ACS is to use the summary files.

Within the Vacancy Status table, there is a count of all vacant housing that are For seasonal, recreational, or occasional use. This probably serves as the starting point for the NAHB analysis, and I wonder if they have additional proprietary data they use to supplement the ACS counts.

You can get the B25004. Vacancy Status data from IPUMS NHGIS. This table is available in the 2020 American Community Survey: 5-Year Data [2016-2020, Block Groups & Larger Areas] (also known as the 2016_2020_ACS5a dataset). If you use the Occupancy/Vacancy and Use option in the TOPICS filter, you should be able to find B25004.

Hi MPC_vanriper,

Thanks a lot for your quick reply!

I suspected they were using that. There does not seem to be yearly ACS data on that though correct? I only found as you point out 5-year data for different periods. I was hoping there was for every year starting in 2000.

Is the idea then to count all housing units that are “For seasonal, recreational, or occasional use” as second homes?

Thanks a lot!

To get data for all counties in the US, you need to use the 5-year ACS summary data. The Bureau uses 5 years of survey responses in order to generate estimates for geographic units with relatively low populations. You can use the 1-year ACS estimates to get estimates for counties with more than 65,000 residents.

The 5-year ACS estimates are available from 2005-09 up through 2016-20. You could consider using non-overlapping 5-year datasets (e.g, 2005-09, 2010-14, 2015-2019) for an analysis if you want to study all counties. Otherwise, you could use the 1-year data to analyze counties with population greater than 65,000.

I assume the NAHB starts with the “For seasonal, recreational, or occasional use” as the basis for estimating second homes. But, they don’t provide enough detail, in my opinion, to know whether that’s all the data they use. I recommend reaching out to them for more details about their methodology.

I just wanted to add that in addition to the ACS data, there are Vacancy Status tables in decennial censuses back to 2000 and beyond, but those, of course, are not available for every year.

Hi MPC_vanriper and Jonathan,

Thanks a lot for the help!

I am interested in looking at all zip code-level data, so as I understand I will need to use the 5-year datasets. Thanks for pointing out that I need to use non-overlapping 5-year datasets.

Best,
Ricardo