Related Topics

New Apartment Construction: How Much Is Needed? (December 2012)


In the wake of the bursting of the housing bubble, the implosion of the financial markets and the deepest downturn of the postwar era, apartment development plunged to the lowest level on record. A strong recovery has brought multifamily construction about halfway back, and it is clearly headed higher. But how high depends on both demand and supply—that is, the number of households that will want to rent apartments in the coming years, as well as the losses to the existing apartment stock and vacancy trends. Assuming current trends hold, over the rest of this decade, we will need at least 300,000 new apartments annually, and possibly as many as 400,000, to meet demand. However, financing constraints, rising construction costs and traditional NIMBY opposition could keep actual construction below that level.


Demand Dynamics: Demographics


Future apartment demand is the outcome of three projections: (1) household growth; (2) the owner-renter split among new households; and (3) the share of new renters who choose apartment living. All three are affected by a variety of demographic, economic and social factors, many of which defy simple modeling. As Nobel physicist Niels Bohr pointed out: “Prediction is very difficult, especially if it is about the future.” For that reason, a forecast range may be of more use than a “single-point” forecast. The chart below shows the recent history of household formations.


2012-12-1
 The decade from 2000–2010 saw the slowest household growth in U.S. history, averaging just 1.0 percent annually. This sluggish growth was largely the effect of the Great Recession, which caused many households to “double up” and led to reduced immigration. (Note that overall population growth also was slow, averaging 0.9 percent annually; but, at that rate, it still was not as slow as in the period from 1930–1940.)

Since economic growth remains measured, one might expect household formation to be tempered going forward, particularly since there is also a natural tendency for population and household growth rates to ease as countries get bigger. However, there is some pent-up demand—households that would have formed earlier but were constrained by economic circumstances—which could add to household formation in the near term. And a stronger economy than is currently expected could increase household growth further. Given these factors, annual household growth seems likely to fall in the range of 1.0 percent to 1.2 percent (the latter figure is on a par with household formations from 1990–2000, when economic growth was strong). This would mean an additional 9.5 million to 12.5 million households from 2013–2020, or 1.2 million to 1.4 million annually. (To be sure, renewed economic weakness could mean household formation below this level.)


Demand Dynamics: Tenure and Structure Type


How will the newly formed households split between owning and renting? According to the decennial census, from 1990–2000, the increase in the number of renters was only 20 percent of the increase in total households. However, from 2000–2010, renters made up 45 percent of net new households—the highest share since 1930–1940, when the country was mired in the Great Depression.


Change in Households (millions)


 


Total


Renters


Renter share


1900


3.3


1.9


59%


1910


4.3


2.4


57%


1920


4.1


2.3


56%


1930


5.6


2.4


43%


1940


5.0


4.3


88%


1950


8.0


-0.4


-5%


1960


10.2


1.0


9%


1970


10.4


3.3


32%


1980


16.9


5.1


30%


1990


11.6


4.3


37%


2000


13.5


2.7


20%


2010


11.2


5.1


45%


Sources: Census Bureau; NMHC


The bursting of the house price bubble led not only to greatly increased foreclosures but also some rethinking of the value of homeownership, at least for the present. As a result, the number of homeowners actually fell slightly in the past five years. With the for-sale market now recovering, that isn’t likely to continue. But the large number of young people now entering the housing market combined with continued tight mortgage credit and households still cautious about the home purchase suggests that the renter share should be higher than in most postwar decades. A renter share of new households ranging from 40 percent to 60 percent is likely. This would put the homeownership rate in 2020 between 63.2 percent and 65.1 percent, compared with the 2012 average rate of 65.5 percent. The increase in the number of all renters, whether in apartments, small multifamily buildings, or single-family houses, would range from 3.8 million to 6.9 million.


Apartments historically have been the residence of choice for about 45 percent of all renter households. In recent years, the apartment share has been lower as a result of the housing bust-induced increase in single-family rentals, as well as the dearth of new apartment supply. Going forward, that share is likely to return to 45 percent. However,  in case that doesn’t happen, and instead the apartment share of new renter households edges down to 40 percent of net new renters, that would mean an additional 1.5 million to 2.8 million apartment renter households over the rest of the decade, or an average net absorption of 190,000–355,000 apartments per year.


Supply Dynamics


By itself, the projected net absorption is insufficient to determine how much new construction is needed; there are additional elements of demand as well as other sources of supply. Specifically:


New construction = net absorption + net losses to the stock + change in the number of vacant units


The apartment stock suffers losses due to destruction (e.g., from hurricanes or fires), deterioration, consolidation (e.g., a renovation might merge two units into one or replace a 250-unit building with 150 units), or net conversion to owner-occupied or nonresidential use (conversions to apartment use are subtracted from conversions from apartments). Data on losses are estimated only indirectly and may be inexact. Nonetheless, the data suggest losses to the apartment stock may reach 75,000 units to 125,000 units each year.


The U.S. Census Bureau estimates there are currently 3.8 million vacant units available for rent. Some level of vacancy is normal and desirable. Assuming the average vacancy rate for the 1980s and 1990s (7.2 percent) is about the right level (note that the average for all years from 1965 to the present is essentially the same at 7.3 percent), then about 3.1 million vacant units would be normal. The excess vacant inventory, therefore, is 700,000 units; this should get worked off as the markets normalize.


Unfortunately, the Census Bureau data do not provide a vacancy breakdown for multifamily buildings with five or more unit rentals. However, assuming the apartment share of vacancies is the same as the apartment share of the rental stock, new apartment renter households would occupy roughly 300,000 excess vacancies over the next eight years—if not sooner than that—at an average annual clip of almost 40,000 units, reducing the need for new construction to that extent.


To be sure, vacancies can also fall below the normal level, so vacancies might fall by more than 40,000 annually. What’s more, some renters may choose currently vacant units in smaller multifamily buildings or single-family homes, if apartment supply remains fairly tight. Assuming these things happen, but only to a limited extent, the numbers suggest that vacant units would meet about 50,000 units of the estimated annual apartment demand.


Conclusion


Where does all this leave us regarding needed new construction of apartments? The projection range is rather wide, the inevitable result of considering different forecast options for several variables. However, it’s likely that the nation will need a minimum of 240,000 new apartments and a maximum of 437,000 units each year from 2013–2020. The table below highlights some of the scenarios.


 New Apartment Construction Needed Annually, 20132020


 


Renter share of net new households


Apt. share of new renters


60%


55%


50%


45%


40%


45%


 


 


 


 


 


Household growth: 1.2%


437,000


405,000


372,000


340,000


308,000


Household growth: 1.0%


370,000


344,000


317,000


290,000


263,000


40%


 


 


 


 


 


Household growth: 1.2%


394,000


365,000


337,000


308,000


279,000


Household growth: 1.0%


335,000


311,000


287,000


263,000


240,000


Sources: Census Bureau; NMHC


For ease of presentation, these scenarios assume 100,000 apartment units are lost each year and that 50,000 vacant units become occupied. These two factors combined increase the needed apartment units by a net 50,000 annually. One could modify this assumption and get an even larger forecast range.


To help gauge what these projections mean, the 2020 homeownership rate in the above scenarios would range from a low of 63.2 percent to a high of 65.1 percent, compared with a 3Q 2012 seasonally adjusted rate of 65.3 percent.


However, this range in annual number of new apartments needed is likely much narrower than the math suggests. As long as there is sufficient new construction, apartments are likely to capture their historical 45 percent share of the net new renter households. Combine that with the more conservative assumptions that the overall renter share of new households will be between 45 percent and 55 percent (that is, leave aside the possibility that it will be either 60 percent or 40 percent). Those conditions produce a forecast range for new apartments needed each year of roughly 290,000 units to 405,000 units (as outlined in the table above), which can be reasonably rounded to a range of 300,000–400,000 units annually.


Questions or comments on Research Notes should be directed to Mark Obrinsky, NMHC’s Senior Vice President of Research and Chief Economist, at mobrinsky@nmhc.org or 202/974-2329.