Multifamily Rental Markets: Growth During the 1990s
Which county added the most multifamily rental units in the decade of the 1990s? Where did the stock of rental apartments grow at the fastest rate? Answering questions like those is one way we can use the data recently released from the 2000 Census.
The Census provides information about occupied rental units, including the type of structure and when it was built. We can use data on structure type to isolate multifamily buildings. We can use data on age of structure to see how many of the occupied apartments were built in the previous 10 years. Although the past isn’t a perfect guide to the future, counties showing a rapid increase in rented apartments can serve as a starting place to look for areas that may continue to be hot markets.
In a typical year, the only construction data with substantial geographic detail that is available from the federal government is the number of building permits issued. (NAHB estimates and forecasts multifamily starts for states and the larger metro areas from permit data.) Permit data suffer from complications, such as: There are areas of the country where construction is not covered by permit; not all permits result in housing starts; and there is a tendency for many multifamily permits to be later reclassified as single family. Moreover, the “lumpy” nature of multifamily construction (that is, many units may be started at once with a relatively long lag between starts) suggests that several years of data need to be accumulated in order to get an accurate picture of the strength of multifamily markets, especially in smaller market areas. But the permit data can at least serve as a rough guide to recent trends in counties that the Census files indicate were “hot” during the 1990s.
A minor drawback to the 2000 Census summary files that have been released so far is that that they don’t identify the vintage of vacant units, so we are restricted to looking at apartments that were occupied at the time of the Census. (In this article we look at renter-occupied units; we will examine condos next month.) An advantage of the Census summary files is that they contain ancillary information about the occupants, and some of this may be of interest — especially data on income and rent paid.
Much of the rental housing stock in the U.S. is single family, and much of the multifamily stock is in buildings with fewer than five units. Although there are exceptions, these smaller structures tend to be mom-and-pop operations, often with owners living on the properties, rather than operations that employ a professional management company. Most multifamily property owners, managers, and builders will therefore be interested primarily in larger structures, so this article focuses exclusively on units in buildings with five or more apartments.
In total, the 2000 Census counted just under 16 million rented 5-plus apartments in the U.S. Unsurprisingly, they tend to be concentrated in certain areas. Among the 3,141 counties (and county equivalents) in the Census, almost one-fourth have fewer than 100 occupied rental apartments in 5-plus structures. Well over half of them have fewer than 500 (Figure 1).
At the other end of the scale, 65 counties have at least 50,000 rented 5-plus apartments, and 11 have more than 200,000. The leader, by a wide margin, is Los Angeles County with about 950,000. This is partly a result of the tendency toward large counties in California, where the entire Los Angeles PMSA (Primary Metropolitan Statistical Area), for instance, consists of a single county. In contrast, the New York PMSA encompasses eight counties, four of which are on the list of those with more than 200,000 5-plus occupied rental units.
Before looking at smaller market areas, Table 1 provides summary statistics for individual states, and for the U.S. as a whole. The number of occupied 5-plus rental units ranges from a low of 14,316 in Wyoming (the state with the smallest population) to a high of 2.4 million in California (the most populous state).
Overall, households renting 5-plus apartments contain 2 persons, on average. Maine shows the smallest average household size (1.59 persons) and California the highest (2.36). The median age of the household head is 39. The “youngest” state is Utah, with a median age of 31; the “oldest” Rhode Island, with a median of 51. For Utah, this 5-plus renter result is consistent with general population statistics, such as the share of the population over or under a certain age, which usually rank Utah first or second to Alaska. It’s perhaps more surprising that Table 1 shows older median age for Rhode Island than for Florida or West Virginia.
West Virginia falls on the bottom of the scale for renters of 5-plus apartments in terms of average household income and gross rent (which includes the cost of all utilities except telecom services) — slightly under $20,000 per year and just over $350 per month, respectively, compared to over $35,000 and $643 for the U.S. as a whole. The possibly unexpected result here is that West Virginia lies below some of the Southeastern states, where median incomes have traditionally been the lowest in the country. A contributing factor is the relatively small average household size of apartment renters in West Virginia.
An indicator of housing market activity frequently used by NAHB is “permit intensity” — the number of permits issued, divided by the population. In Table 1, the permit intensity measure reported is the number of units authorized by permits in structures with five or more apartments in 2002 per 100,000 residents. Because the Census Bureau has not yet released annual permit figures for 2002, the intensity numbers reported in the table are based on aggregated monthly data. When they are released, the final 2002 permit totals will differ from this, but the discrepancies are usually not large.
For the U.S. as a whole, 2002 intensity was 119 5-plus permits per 100,000 people, but this varied considerably from state to state. Nevada ranked first at 389, and Florida also was over 300.
Most New Construction
Typical housing markets are much smaller than states, so we next consider county-level Census data. Counties are often used as a unit of analysis simply because little data is available below the county level. In project feasibility studies, market areas are typically defined by a radius. The radius varies from project to project, but can encompass an area as small as 80 square miles or as large as 2,800 square miles. Excluding the anomalous boroughs and Census-designated areas of Alaska (which are equivalent to counties in the Census data), more than 95% of the counties have land areas that fall within those extremes.
One way to measure the multifamily rental markets that expanded the most during the 1990s is simply to look at how many occupied 5-plus rental units were built during that time. Table 2 shows the top 10 counties on this scale.
The counties in Table 2 all contain central cities with a population of at least 1.5 million (Harris County in Texas, for example, contains the city of Houston) and most are considerably larger than that. The smallest, in terms of population, is the Las Vegas metro area which includes Clark County, NV. That county stands out by having the fastest rate of increase of any county in the table, having built more than 40% of these units in the previous decade.
Permit intensity shows that, at least for these large counties, the trends of the 1990s generally continued into 2002. The 2002 permit intensity column is highly correlated with share built 1990-2002 column. Clark County is first by a substantial margin in both cases. The permit data suggest that activity has picked up recently in the Miami area, however.
Fastest Growth Rates
Table 2 captures counties that had a large number of multifamily units built during the decade of the 1990s. These tend to be counties that began the decade with a large stock of multifamily units. Thus, Los Angeles County heads the list, even though less than 10% of its renter-occupied 5-plus housing stock was built in the 1990s.
Another way to look at housing markets is in terms of growth rate. Which counties built a particularly large fraction of their renter-occupied multifamily stock after 1989? Table 3 shows the top 10 counties on this scale.
Because we don’t want to consider very small levels of production, the table is based only on counties that had at least 500 renter-occupied 5-plus apartments in 2000. The shares built in the 1990s generally are quite high, and the table essentially shows counties that developed multifamily rental markets in the 1990s where they didn’t exist before. Henry County, on the southeastern edge of the Atlanta metro area in Georgia, is at the top of the list with well over 80% of its rented 5-plus units built in the 1990s.
The table also shows average renter income and gross rents paid. There is a considerable spread, from under $25,000 and $444 in Christian County, MO to over $50,000 and $900 in Douglas County, CO and Summit County, UT.
Permit intensity does not indicate that the markets in Table 3 continued to be hot after 2000. Although the intensity measure is high for the first two counties on the list (Henry and Douglas), it is below the national average for the others, and zero (meaning that no 5-plus permits were issued in 2002) in three cases. The lumpy nature of multifamily production might make one-year intensity an unreliable indicator in these relatively small counties, however. Average multifamily construction activity can be high over a period of several years, while few (or even no permits) are issued in a particular year, due simply to the timing of the start of large projects.
For this reason, we might like to consider counties that represent somewhat larger multifamily markets. Table 4 is similar to Table 3, but is restricted to counties that had at least 5,000 rented 5-plus units in 2000 (fewer than 400 counties, in total).
The counties that make this table generally belong to metropolitan areas, but do not contain the area’s primary central city. Starting from larger bases, the counties in Table 4 show somewhat slower growth rates than those in Table 3 during the 1990s. Collin County, on the northern part of the Dallas metro area, ranks first, reporting that it built 58.7% of its 5-plus rented units (nearly 23,000 apartments) over the decade.
Permit intensity suggests that apartment construction activity cooled in Collin County — as well as in Gwinnett County (Atlanta metro area), Fort Bend County (Houston metro area), and Washington County (Portland, Oregon metro area) — after 2000. But the other counties in Table 4 seem to have remained quite hot, or even to have become hotter. Permit intensity in 2002 was as high as a scorching 919 in Osceola County, FL (Orlando metro area).
NAHB has compiled a complete set of tables containing detailed Census information for 5-plus apartments in every county in the United States. The information includes numbers of both rental apartments and condos, when they were built, when occupants moved in, and their characteristics such as age, household size, and average income by age of the household. For rented units, the tables show average gross rent and rent burden (gross rent as a percentage of household income). For condos, the tables show the average market value of the units. This information, customized for a particular county, or for several counties, is available for a fee. Contact firstname.lastname@example.org for more information.
return to top ]