This report is courtesy of the National Multi Housing Council.

In the aftermath of the Great Recession, absorptions dried up, causing the number of vacant apartments to rise beyond the normal range and creating what are termed “excess” vacant units. This excess inventory signaled a supply-demand imbalance that led to a sharp cutback in new production and lower returns. Over the past three years, however, the increase in apartment demand has outstripped the increase in supply, causing occupancy rates to trend up again and reducing excess apartment inventory. But to what extent has that excess inventory of vacant apartments shrunk?

Unfortunately, public data are too limited to provide a clear and simple answer. What follows is a brief summary of data sources and issues, along with a standard estimate of excess apartment vacancies. Following that is an assessment of the reliability of this sort of estimate. Our best estimate is that there is little excess vacancy nationally. This means that while there is excess vacancy in some metro areas, it is offset by other markets where the number of vacant apartments is below average.

Overview of Public Data on Vacancies

There are four main federal data sources that provide estimates of renter vacancy rates: Housing Vacancy Survey, Decennial Census, American Housing Survey and American Community Survey.

The most widely used is the Housing Vacancy Survey (HVS). It is available quarterly, making it the most frequent and timely data source, and has metro area vacancy rate estimates as well as vacancy rate estimates specifically for apartments. The sample size is rather small, however, and it provides housing inventory data only for all renters combined, with no breakdown by housing type, such as single-family rentals, small multifamily buildings (2-4 units per building) or apartment buildings (5+ units).

The Decennial Census has the most reliable data as it covers all households and housing units and has the most extensive geographic breakdowns. How-ever, the drawbacks are that the data are available only every 10 years and no longer include a breakout for apartments.

The American Housing Survey (AHS) is available more frequently—every other year—and includes detail on the number of units in the housing structure. However, similar to the HVS, it has a fairly small sample size. Metro area data also are not generally available, and users must dig into the micro-data files to access relevant information.

By comparison, the American Community Survey (ACS) has a much larger sample, metro area renter vacancy rates and a detailed breakdown by units in structure (national data only, not by metro area). However, similar to the AHS, users must manipulate the microdata files for the data, which greatly restricts usage of this dataset.

Determining the Right Level of Vacancy

The optimal vacancy rate balances the desire of prospective residents to find homes against owners’ desire to manage the business in an efficient and cost-effective manner. While an exact measure may be hard to ascertain, the long-term average provides a reasonable approximation of a “normal” vacancy rate, which will suffice for present purposes.

The preceding chart shows a rising trend for the overall rental vacancy rate, which includes single-family and small multifamily rentals as well as apartments. By contrast, the rate for just the apartment sector trended down in the 1990s before turning up again in the current decade, first as a result of the housing bubble, then as a result of it bursting (and taking the economy with it). In estimating a normal rate, based on historical trends, it probably makes sense to exclude the bubble period (2000 and after).

From 1970-1999, the overall rental vacancy averaged 6.5 percent, while the vacancy rate for 5+ multifamily units averaged 8.7 percent. (Note that Census Bureau vacancy rates tend to be higher than those reported for investment-grade apartments.) This means that the overall rental vacancy rate of 8.7 percent in 2012 was about 220 basis points (bps) over the normal rate, while the apartment vacancy rate of 9.3 percent was only 60 bps over the normal rate.

Estimating the Vacant Rental Inventory

The table below summarizes the rental housing inventory for 2012 and also what that inventory would look like if the overall renter vacancy rate was at the normal level.

With a normal (6.5 percent) vacancy rate, roughly 2.8 million rental units out of a total of 43.9 million would be vacant. At the actual vacancy rate, there were 965,000 additional, or excess, vacant units. In other words, almost one million additional rental units would need to be absorbed (with no new production) to get the vacancy rate back to its normal level.

How much of that excess inventory is in the apartment sector? Unfortunately, the HVS does not provide the detailed breakdown on total, occupied or vacant units by units in structure—that is, no detail on the single-family, 2-4 unit multifamily or 5+ unit multifamily stock.

Such detailed estimates are available in the micro-data from the AHS and ACS, although the most recent data are from 2011. The former pegs the total 5+ unit multifamily rental stock, or apartments, at 18.1 million units; the latter estimates a total 19.4 million units. Since the apartment renter vacancy rate in 2012 was only 60 bps higher than the normal rate, we can simply apply that figure to the units in the stock to get an estimate of the excess vacancies in the apartment world. Using the AHS, we get 110,000, while ACS data yields a figure of 118,000. These are both very small figures, given the overall size of the apartment stock.

However, some caveats are in order. First, the small sample sizes in the HVS and AHS mean large confidence intervals around these estimates; in other words, they shouldn’t be construed as exact. Second, this method requires mixing estimates from different surveys. Since the surveys have different survey samples, sample sizes, interview procedures and weighting schemes, researchers generally avoid this method of estimation, if possible.

More fundamentally, these surveys also provide somewhat conflicting estimates on rental vacancies and the rental stock. A brief excerpt from the data-sets illustrates the problem.

According to the HVS there are 24 percent more vacant rental units than in the ACS, while the total rental stock in the ACS is a bit higher (44.5 million units) than in the HVS (43.0 million units). This produces quite different vacancy rates: 9.5 percent (HVS) versus 7.4 percent (ACS). The 5+ unit renter vacancy rate is similarly higher: 10.4 percent (HVS) versus 8.6 percent (ACS). While the AHS and HVS estimates are much closer, that doesn’t mean they are more likely to be correct; the ACS has a much larger sample and also better matches the data from the Decennial Census. With no compelling case to use just one of these datasets, we are left having to consider them all. The Census Bureau is aware of the problem with widely differing estimates for key housing data and is working to find a resolution.

Even with these cautions, these data are strong enough to support our conclusion: There is little to no excess apartment vacancy today.

________________________________________

Questions or comments on Research Notes should be directed to Mark Obrinsky, NMHC’s Vice President of Research and Chief Economist, at This email address is being protected from spambots. You need JavaScript enabled to view it. or 202/974-2329.