The rate of housing removals is an often overlooked, but surprisingly important component of housing demand. It is often overlooked because quality data on home removals are scarce to nonexistent. It is important because removals account for about 20% of the demand for housing over the course of a housing cycle. Only net household formation accounts for a bigger share of underlying demand for housing.
Unfortunately, none of the big government statistical agencies publish data on annual housing removals. Every now and then, a private group will take a stab at providing a point estimate in a given year. But these estimates usually only consider a subset of removals. For example, the National Association of Homebuilders recently estimated that nearly 60,000 homes were removed in 2017 to make way for new single-family homes. While this data point is interesting, it does not give us a complete picture of the removal category as it leaves out the multifamily and mobile home sectors and does not account for homes that are removed from the housing stock for other reasons.
The U.S. Census Bureau does not publish an estimate of annual housing removals, but it does provide the tools for motivated analysts to generate these numbers themselves. Census researchers publish housing unit loss rates in the methodology document found within the Population and Housing Units Estimates section of the Census website. These loss rates are updated on a sporadic basis, and the last update took place in 2011. The data for housing loss are taken from the biannual American Housing Survey. The following situations count as a permanent loss of a housing unit according to this survey:
1. Interior exposed to elements
2. Unit demolished or lost in a disaster
3. House or mobile home is moved
A housing unit is considered lost if it appeared in 2009 but was included in one of the three categories above in 2011. The loss rates vary by the decade in which the housing unit was built. The survey considers single and multi-family housing units, mobile homes and other types of homes.
The most recent loss rates are seen in Table 1. Note that homes built after 2000 are assumed to have a loss rate of zero and that all homes built before 1940 have the same loss rate whether they were built in 1939 or 1899. These factors would tend to bias our overall housing removal estimate downward somewhat.
To calculate our housing removals estimate, we apply various loss rates to the housing stock broken down by the year the structure was built. Fortunately, that data is easily found in the 2015 American Housing Survey (which was released in 2017). This data for single and multifamily housing is shown in Graph 1.
After applying the appropriate loss rates to the housing stock figures in shown in Graph 1, we get an annual housing removals figure of 200,000. Next, we consider mobile homes, of which there are relatively few, but which have a high loss rate of 1.76%. The mobile home loss rate is not broken out by the year in which the homes were built. The AHS Survey indicates that there were 8.686 million mobile homes in the U.S. in 2015. Applying the loss rate to this figure translates to a removal figure of 153,000 for this segment. Mobile home shipments have averaged only about 80,000 units per year over the last three years, which means that the installed base of mobile homes is shrinking in the U.S. Mobile homes are likely to continue to lose market share over the next 10- 15 years as their popularity continues to wane.
The most reliable data on population and household formation are only released once every decade when the Decennial Census is released. As a result, the unit of observation for our underlying demand estimate is a decade rather than a year. We typically update our estimates for other determinants (e.g., underlying housing demand such as removals and second home demand) on an intermittent basis as data quality permits.
For the 2010-2019 period, we have raised our estimate for annual housing removals from 300,000 per year to 350,000 (200,000 single and multi-family units plus 150,000 mobile homes). We are also assuming that housing removals in the coming years will average 350,000 annually from 2020 to 2029, but we caution that this figure could be to low due to the aging of the housing stock. After all, the median housing unit in the U.S. was built in 1976.
Bottom line: Housing removals comprise a significant portion of housing demand, and there is a paucity of data for this key indicator. Using loss rates calculated by the Census Bureau and applying them to the U.S. stock of housing units and mobile homes allows us to calculate a plausible estimate of 350,000 housing removals per year in the U.S. for the 2010-2019 decade.
This article was excerpted with permission from FEA’s “Spotlight.” To learn more, visit www.getfea.com.