Until recently, the apartment sector was praised for showing remarkable discipline in limiting new construction. Now, with higher vacancy rates and no significant drop in construction, many have concluded that the industry is again overbuilding.

Recent National Multi Housing Council (NMHC) research examined what exactly is the “right” level of building. Our analysis concludes that current construction levels, when viewed as part of the big picture, may indeed be appropriate.

Over the past 35 years, the apartment industry has recorded enormous swings in construction, with production more than doubling and then falling by 50% to 80%. The biggest wave of apartment construction occurred in the early 1970s in response to a strong economy, increased demand from the first wave of baby boomers reaching adulthood, government “urban renewal” incentive programs and easy credit. The biggest decline followed right on its heels, however, thanks to a recession and substantial reductions in government subsidy programs.

Construction picked up again in the second half of the 1970s and then declined in the early 1980s. Then changes in federal tax law in 1986 caused construction to plummet until the mid-1990s.

Apartment Stocks and Flows
(units in thousands)
Decennial Census American Housing Survey-1 American Housing Survey-2
Stock, 2000 17,252 15,632 15,936
Stock, 1990 15,898 15,100 14,928
Net change 1,354 532 1,008
Completions 1,820 1,796 1,865
Implied net loss 466 1,264 857
Annual loss rate 0.3% 0.8% 0.6%
Note: The column labeled AHS-1 compares the apartment stock in 1999 and 1989, while AHS-2 compares the apartment stock in 2001 and 1991
Source: Census Bureau; NMHC tabulations of microdata from the 1990 and 2000 decennial census public use microdata sample (PUMS) files; NMHC tabulations of microdata from the AHS for 1989, 1991, 1999, 2001.


These wide swings no doubt entailed considerable costs: apartment firms first obtain and then discard permits, labor, financing and materials. Since these cost “spikes” are ultimately borne by consumers, the steadier construction levels of the 1990s is surely more desirable.

Multifamily Arithmetic

Approximately 250,000 to 300,000 new multifamily units have been built annually in recent years. But for the past seven years, only 155,000 of those, on average, are actually new, unfurnished, market-rate rental units.

The rest consist of condos and co-ops, timeshares and other housing not intended for the rental market. While these units may, to some extent, compete with market-rate apartments, they constitute distinctly different markets.

Many new apartments are merely replacing units lost to physical deterioration or conversion to non-residential use. Unfortunately, there are no direct surveys that measure apartment stock losses. But they can be indirectly measured by estimating the stock at two points in time and comparing that to the number of new units constructed between those points in time. The difference between the “flow” of new units and the net change in the stock provides a measure of units lost.

The decennial census says the apartment stock increased by 1.4 million units — to 17.3 million units — between 1990 and 2000. Part of that increase, however, represents units not counted in 1990, but counted in 2000. (The 1990 census missed more people and households than the 2000 census.)

The American Housing Survey (AHS) provides another frame of reference. According to the AHS, which is only available in odd-numbered years, the stock increased by 500,000 between 1989 and 1999 and by 1 million between 1991 and 2001. Subtracting this from the 1.8 million new apartments completed during the relevant 10-year periods produces an implied net loss that ranges from 466,000 to almost 1.3 million.

On a percentage basis, this means the nation loses an estimated 0.3% to 0.8% of its apartment stock each year. The known problem of better counting in the 2000 census suggests that the loss rate estimate of 0.3% is probably too low. The true loss rate is probably between 0.5% and 0.8%.

Adding It All Up

Annual completions of market-rate apartments have averaged about 155,000 over the past seven years, but by applying the estimates above, we can conclude that approximately 50,000 to 80,000 units are lost annually by multiplying the loss rate (0.5% to 0.8%) by the estimated 10 million market-rate apartments.

That means the industry is adding no more than 75,000 to 105,000 new market-rate rental units each year. Stated alternatively, it is growing approximately 0.8% to 1% annually. Given the wave of immigrants and echo boomers expected to flow into the rental housing market in the next decade, this level of construction seems just slightly below the likely increase in demand.

To be sure, in periods when net absorption is negative, any increase seems too much. And some markets have no doubt seen building in excess of the likely demand increase in those areas. But overall, the level of new construction seems to be reasonably well-suited to demand over the intermediate- to long-term.

Mark Obrinsky is vice president of research and chief economist of the Washington, D.C.-based National Multi Housing Council.