Special Studies, October 7, 2010
By Paul Emrath, Ph.D.
Vice President for Survey and Housing Policy Research
National Association of Home Builders
Report available to the public as a courtesy of HousingEconomics.com
Each year, the Census Bureau’s Manufacturing and Construction Division publishes tables showing various characteristics of new housing.  The report is based on data collected in the Survey of Construction (SOC), which is also the instrument used to generate the familiar series on housing starts.
The tables published on the Census Bureau’s website show characteristics of homes completed during a given year. For some purposes, it may be more convenient to look at characteristics of homes started. “Starts” has become a more widely recognized term, as well as the one most often reported in the media as a measure of new construction. Also, completions follow starts with a lag, so data based on starts can provide a somewhat more current picture of changes that may be occurring in the marketplace.
In 2009, the SOC began collecting additional information on new housing, based in part on suggestions from private- and public-sector stakeholders, including NAHB. There is a new SOC question on age restriction, for example, that was strongly advocated by NAHB. Although these new characteristics have not yet begun to appear on published tables, the Census Bureau has made them available in a data set that researchers can download and tabulate.
This article is based on tabulations of the publicly available SOC data set and shows characteristics of new single-family homes started in 2009,  with an emphasis on housing characteristics and geographic detail not shown in standard tables published on the Census Bureau’s website.
Before presenting data on characteristics available for the first time in 2009, Table 1 shows how a standard set of standard SOC characteristics changed over the 2005-2009 period. Perhaps the most obvious trend over this period is the sharp decline in the number of single-family homes started, from 1.7 million in 2005 all the way down to under half a million in 2009.During that time the share of “spec” homes (those built for sale, typically in relatively new residential subdivisions) also declined from nearly 80 percent to less than two-thirds, illustrating the way spec homes were particularly hard hit during the downturn.
Another trend that started to develop in 2007 is a decline in the median size of single-family homes started (which doesn’t show up in the annual series for new homes completed until 2008). From a peak of 2,268 square feet in 2006, the median size of new single-family homes started dropped consistently down to an even 2,100 in 2009. From a historical perspective, the last time the size of new homes declined like this coincided with the recession of the early 1980s. That time, the decline turned out to be temporary. This time, part of the current home size decline may again be a temporary recession-related phenomenon, but part can also be attributed to trends in factors like the desire to keep energy costs down, amounts of equity in existing homes available to roll into a new one, tightening credit standards, less emphasis on the pure investment motive for buying a home, and an increased share of homes sold to first-time buyers.Not all of these trends are likely to reverse themselves immediately at the end of a recession.
The decline in the median size of new homes even managed to occur as the share of homes that were contractor-built—typically the category with the largest homes on average—was increasing. Meanwhile, the median sales price of single-family homes started has also declined—from $256,000 in 2006 to $211,000 in 2009. For purposes of the SOC, the Census Bureau defines sales price as the price agreed upon between the buyer and seller in the initial contract and applies only to homes that were actually sold. The equivalent measure for contractor-built homes is median contract price, which has followed a more erratic pattern since 2005. Contract price is defined as the price on the initial contract between the owner and general contractor and therefore excludes the value of the land, subsequent change orders, and any work that may be performed on the lot or house by other contractors. The share of work done by the main contractor on a custom home could vary over time and contribute to the fluctuations in median contract price shown in the table.
While both median size and median sales price have been declining, the average numbers of bedrooms and bathrooms per house have shown little change. The downward trends in house price and size also show up as a declining share of homes built at the high end of the scale, particularly in the later years. For example, the share of for-sale homes priced above $300,000 was over 35 percent in 2006 and 2007, before dropping to under 30 percent in 2008, and under 25 percent in 2009. Similarly, over 9 percent of the single-family homes started in 2007 and 2008 were 4,000 square feet or larger, compared only 7.3 percent in 2009.
Among the various amenities captured in the SOC, three-car garages, fireplaces, patios, and decks have each declined as a share of single-family starts since 2005. These amenities could be viewed as luxuries, and their declining shares therefore consistent with the reduced share of homes being built at the top end of the market. On the other hand, the share of new single-family homes with porches has risen consistently since 2005.
The incidence of heat pumps in new homes has also been on the rise, which should not be surprising, given the recent focus on energy efficiency and efforts by the Department of Energy and the Environmental Protection Agency to promote certain types of heat pumps for residential use.
A noticeable trend in siding material is the declining share of new single-family homes built with stucco as the primary siding—likely the result of a geographic shift in construction activity. The share of new single-family homes started the West, where stucco is traditionally most common, declined from roughly 25 to 20 percent of the total market between 2005 and 2009.
The standard tables published on the Census website show characteristics of new housing for each of the four principal census regions. By tabulating the public use SOC data set, it’s possible to show characteristics of single-family homes started at the slightly more detailed geography of the nine census divisions (Figure 1).
The last 9 columns of Table 1 break down the standard characteristics discussed in the preceding section by division. The aforementioned geographic pattern to the use of siding material is evident in this section of the table. Stucco tends to be the most common (primary) siding material in the Mountain and Pacific divisions, brick in the two south central divisions, and vinyl has the lion’s share of the market elsewhere.
Table 1 also shows that decks tend to be particularly uncommon on single-family homes built in the West South Central Division—the division with the second highest incidence of patios. Across all nine divisions there is a strong negative correlation between the shares of homes built with patios and decks, suggesting that builders tend to see patios and decks as substitutes for each other. Builders responding to the SOC reported that they provided both a patio and a deck on fewer than 6 percent of new single-family homes started in 2009.
There is no similarly strong correlation between porches on the one hand, and either patios or decks on the other—suggesting that a porch is a design element of the house that doesn’t often influence the decision to provide a patio or deck.
There is also a geographic pattern to the share of new single-family homes built for sale, rather than on an individual customer’s lot. In 2009, the share of homes built for sale was well over 70 percent in the South Atlantic, Mountain, and Pacific divisions, but under 45 percent in New England and the East North Central. The column for the New England Division in Table 1 should be interpreted with caution, however, as that division has the smallest number of single-family homes started, producing statistics with the largest margins of error.
Some of the divisional differences in amenities shown in Table 1 can be attributed to climate. For instance, air conditioning is less of a necessity in cooler parts of the country. The most common variety of heat pump technology tends to be more efficient in parts of the country where the temperature stays above freezing most of the time. Basements may be more difficult to build in certain areas due to soil conditions, but building codes also require deeper footings in cooler climates, reducing the cost difference between a full basement and shallower type of foundation.
A First Look at New Characteristics
In terms of physical characteristics of the house, the SOC began collecting information on two new items in 2009: presence of a two-story foyer and location of laundry connections. Overall, 35 percent of single-family homes started in 2009 had two-story foyers. This tends to be a luxury feature, however—considerably more common in more expensive homes (Figure 2).
The 2009 SOC data show that almost all new single-family homes have washer and dryer connections, and almost always confined to a single location in the home. Nearly 80 percent of the time the single washer-dryer location is on the first floor. Figure 3 shows how use of two relatively uncommon laundry locations—the basement and the garage—varies with size of the home. Not surprisingly, builders resort to these non-standard locations for laundry connections more often in smaller homes, where competition from alternate uses for first-floor house space is especially strong.
In 2009, the SOC also started collecting data on water and sewer connections. Nearly 80 percent of new single-family homes started in 2009 were connected to a public sewer system, and more than 87 percent were connected to a public water supply. Other types of water and sewer systems (primarily individual wells and septic systems) are used more often in custom homes (i.e., homes built on the customer’s lot whether the customer acts as the general contractor or hires a builder) than on homes built in tracts for sale (Figure 4).
Other data collected by the SOC for the first time in 2009 include additional information on the legal structure of communities in which the single-family homes are being built. Prior to 2009, the SOC collected information about condominiums, but legal community organizations also exist under different names, such as homeowner’s association. The functions of these associations may be quite varied and include any combination of fee collection, restrictions on property use, and access to shared facilities.
Over 47 percent of the new single-family homes started in 2009 were in communities governed by a homeowner’s or other type of community association, but there was a distinct geographic pattern to this phenomenon. It was most common for new homes to be built in association-governed communities in the Mountain and South Atlantic census divisions (Figure 5).
Some associations are set up partly in order to control the age of eligible residents. According to Fair Housing law, it is possible to legally restrict the age of occupants in a new development under one of three sets of conditions. In a single-family development, the set of age-restricting conditions most likely to be used require that the community be able to demonstrate an intent to provide housing for persons age 55 or older, and that at least 80 percent of the occupied housing units in the community contain at least one occupant of this age.
Reasons a developer may choose to establish an age restricted community include relief from school or other impact fees, reluctance of local jurisdictions to approve other types of projects, and guaranteeing that a community with amenities that may attract older residents is not the target of Fair Housing litigation.
Overall, 2 percent of single-family homes started in 2009 were in age-restricted communities, but this is another phenomenon that exhibits a clear geographic pattern. Age restriction was most common in new single-family construction on the east coast in 2009, in either the Middle Atlantic or South Atlantic census divisions (Figure 6).
Another tendency evident in the data is the scarcity of age restricted single-family housing at the extremes of the price distribution. In the 2009 SOC sample, none of the new single-family homes priced below $100,000 or above $1,000,000 were started in age-restricted developments. This doesn’t mean that precisely zero age-restricted homes were started in these price ranges, but that age-restriction is rare enough at these prices to make detecting it difficult with a sample the size of the SOC. In general, age restriction was most common for single-family homes priced between $250,000 and $500,000 (Figure 7).
More detailed tabulations of the characteristics introduced into the SOC in 2009 are available in a file that can be opened or downloaded from the “other resources” box that appears at the top of this article when viewed on NAHB’s website. The downloadable file shows the new SOC characteristics for single-family homes started in 2009 broken down by construction method (site-built, modular, or panelized), census division, house size, sales category (for-sale, contractor- or owner-built), metropolitan and age-restricted status, and sales price.
SOC New Characteristics Table (Excel)
See other Special Studies
Although the SOC collects data on multifamily as well as single-family construction, the public use data set excludes multifamily structures due to confidentiality concerns (it can be relatively easy to identify a specific multifamily property from its basic characteristics, which would violate Census disclosure rules).
 See NAHB’s basic forecast of housing starts beyond 2009. More detailed information, including forecasts of housing activity by state and metro area is available to subscribers of HousingEconomics.com.
 For information on the share of first-time buyers in the market, see the September Special Study on “Characteristics of New and First-Time Home Buyers”
 This is about half the percentage you would expect if decks and patios were independent of each other, rather than functioning as substitutes, based on the numbers shown in Table 1.