How Features and Location Affect Rents
The tricky art of pricing apartments is the focus of NAHB’s newest statistical model. Designed to show how different amenities and features can impact gross rent, the model can help apartment developers and owners place a value on these features as they compare rents in their region.1

The model also can be used to determine if and how particular property renovations would pay off for owners wanting to increase the rent. And property owners and local governments could use it to study how neighborhood characteristics affect rents in nearby apartment buildings.

The model was developed from data from the most recent (2003) American Housing Survey (AHS, conducted by the U.S. Census Bureau and the Department of Housing and Urban Development)².   The AHS contains considerable information about rental apartments and the neighborhoods in which they're located. Although it includes monthly contract rent as well as gross rent (which includes contract rent, plus payments for all utilities except the telephone), the model is based on gross rent. Gross rent generally is considered more reliable for this purpose, due to uncertainties about how the utility costs may or may not be incorporated into the contract rent reported to the Census Bureau. 

The amount of geographic detail in the AHS is somewhat limited, so the NAHB model estimates an average gross rent across a broad Census region—not the rent for a particular apartment in a specific neighborhood.  Also, no survey or statistical model can possibly capture all features that potentially affect market rents, so there's always a chance that a particular feature in the model is acting partly as a proxy for others. If larger apartments tend to have higher-quality appliances (something no one can tell from the data), for example, the estimated impact of the apartment’s size could be picking up part of the effect of upgraded appliances—as well as the value of the extra square footage itself.

Subject to these caveats and limitations, the model may be used to show how changing the age, structural features, general location, and neighborhood characteristics of an apartment in a building with five or more units tends to affect its market rent. Thus, builders and prospective renters can use the model  to get an idea of how the addition of a particular amenity would affect the rents charged and paid for a similar apartments in the region.   Developers looking to expand into other areas can use it to compare rents for a particular type of apartment across broad geographical regions. Owners of rental properties also can use it to judge how a particular renovation would tend to increase the rent they could charge. And property owners and local governments could use it to study how neighborhood characteristics affect rents in nearby apartment buildings.

This article illustrates some of the potential uses of the model by analyzing a benchmark or “standard” apartment, defined as one with the following features:
• 1,000 square feet of living space
• 2 full bathrooms (no half bath)
• 2 bedrooms
• 2 miscellaneous rooms (i.e., rooms other than bed or bath)
• located in a 3-story building
• no extra amenities (such as a balcony, fireplace, elevator, building security system, or use of a garage).

These characteristics are based on median values and percentages for the AHS sample of rental apartments built after 1999. 

Location, Location, Location

Economic theory suggests that an apartment’s location will have an impact on its market rent. On the demand side, tenants may be willing to pay more for a particular location—one near a major employment center, for example. On the supply side, the amount of developable land varies with location, as do certain construction and development costs.

The AHS identifies the four principal census regions and the urban status (central city, suburb, and non-metro) of the area in which a home is located, but not the specific state or city. Some metro areas are identified, but there are generally too few observations to treat each metro separately in the model. It was possible, however, to carve out a number of the larger California metros—where prices tend to be higher than in the rest of the West—and treat them as a separate "region."  Figure 1 shows the estimated price of a standard new home built in each of these areas.

Source: NAHB hedonic regression model for rental apartments in buildings with five or more units; based on data from the  American Housing Survey, U.S. Census Bureau and the Department of Housing and Urban Development

There are relatively few large apartment buildings outside of metropolitan areas, making it difficult to distinguish regional differences in non-metro areas, so there is only one estimated gross rent for non-metropolitan areas, irrespective of which Census region it's located in. 

Except in the West region (exclusive of the large California metros), rents are higher in suburbs than central cities, and higher in central cities than non-metro areas. The model estimates that gross rent for the standard apartment will range from $623 (if it’s built in a non-metropolitan area) to $1,125 (if it’s built in a suburb of one of the large California metro areas).

The statistical model found a strong relationship between rents and the number of stories in the building. Rents tend to be higher in taller buildings (Figure 2).  Although it’s possible this is revealing something about renter preferences, it’s more likely that building height is serving as a proxy for location. Where land prices are high, it tends to drive land toward more intense uses, resulting in taller structures.

Source: NAHB hedonic regression model for rental apartments in buildings with five or more units; based on data from the  American Housing Survey, U.S. Census Bureau and the Department of Housing and Urban Development.

The estimated gross rent for a standard apartment in a southern suburb ranges from $694 if (it’s located in a one- or two-story building) to $1,223 (if it’s located in a building with 10 or more stories).  A southern suburb is used for purposes of illustration, because that’s where the largest share of new multifamily construction is taking place.  

Part of the reason for the relatively high rent in the 10-story-or-higher building is that we are more or less forced to assume in that case that the building has an elevator. The data indicate that there are virtually no apartment buildings of that height in the U.S. without working elevators. On the other hand, there are almost no two-story multifamily buildings that do have elevators, and elevators are comparatively rare in three-story buildings. So the estimates in Figure 2 are based on buildings without elevators, except for the tallest category of buildings.³  The rent difference between apartments in four-to-nine and the 10-story apartment therefore reflects the value of an elevator, as well as possible differences in land prices.

Age Makes a Difference

It’s also possible to look at the price of a standard apartment based on when it was built.  Figure 2 does this for a standard apartment in a southern suburb. There is a clear tendency toward higher rents in newer apartments, which is consistent with results found in many other studies. The effect persists after controlling for many factors with a statistical model. Thus, the rent differences in Figure 3 are for a standard apartment of the same size, with many of the same features, and in the same location.

Source: NAHB hedonic regression model for rental apartments in buildings with five or more units; based on data from the  American Housing Survey, U.S. Census Bureau and the Department of Housing and Urban Development.

Why do newer apartments command higher prices even after controlling for other factors?  In part, age of the structure may be acting as a proxy for features not available in the data.  Newer apartments may have floor plans—or be in properties with common areas—that are more in keeping with current tastes. Newer apartments also are usually built to more stringent codes and are free of  some perceived safety and health hazards such as asbestos and lead-based paint. In some cases, development costs for newer buildings may be higher,  if local jurisdictions have increased construction-related fees over time.

In addition, maintenance costs are likely to be lower for units in newer buildings. NAHB has shown that this is true for owner-occupied housing units,⁴  but equivalent data for rental properties are hard to come by. Private data on operating costs exist, but these typically do not include age of the structure. NAHB has been working closely with HUD and the Census Bureau to collect more information about rental properties in the Residential Finance Survey, however. When those data become available, they should allow us to say more about costs and the age (and other characteristics) of multifamily rental properties. An article on the subject is planned for a future issue of Multifamily Market Outlook.

If it does, in fact, cost less to maintain newer properties, owners may be using the savings to provide more attractive common areas, or other amenities not captured in the AHS.

Extra Features Cost More 

An interesting aspect of the model is the way it can be used to study rents when features of the apartment change. This feature of the model allows builders and prospective renters to use it to get an idea of how a particular amenity affects the rents charged and paid for similar apartments in the region. Builders can use it to help determine if the cost of providing a particular amenity is worthwhile. Households contemplating a different apartment can use it as a preliminary search tool, to get a rough idea of the cost of different amenity packages. Owners of rental properties also could use it to help judge how a particular renovation would tend to increase the rent they can charge.⁵

Figure 4 shows how the estimated gross rent of a standard new apartment built in a southern suburb responds to changes in its features. The feature with the largest impact on rent is a full bathroom. The rent difference between otherwise similar one and two bedroom apartments in Figure 4 is $88 a month.

Source: NAHB hedonic regression model for rental apartments in buildings with five or more units; based on data from the  American Housing Survey, U.S. Census Bureau and the Department of Housing and Urban Development.

On the other hand, an extra miscellaneous room (that is one other than a bedroom or bathroom) has a relatively small impact of $11 per month. It’s important to remember that this number assumes that other factors, including the total square footage of the unit, are held constant. 

What's in the Neighborhood?

The NAHB model also controls for attributes of the neighborhood in which the apartment is located. As result, developers can use the model to help evaluate potential building sites.  Local governments and community organizations also could use it to estimate the impact of certain public policies (such as providing public transportation, or finding a use for abandoned buildings) on home values in their neighborhoods.

Figure 5 illustrates the impact of several neighborhood features, again using a standard new apartment built in a southern suburb as the starting point. The figure shows that the features with the largest positive impact on rents are recreational amenities. Locating an apartment in a community with a clubhouse or walking trails—or near a body of water such as a lake, river or ocean—adds $42 a month to the rent. 

Source: NAHB hedonic regression model for rental apartments in buildings with five or more units; based on data from the  American Housing Survey, U.S. Census Bureau and the Department of Housing and Urban Development.

Neighborhood features may also be “disamenities,” in that their presence reduces the rents people are willing to pay for an apartment. The AHS provides information on several potential disamenities. Among them, one with the greatest negative impact on rents is the presence of either abandoned buildings or buildings with metal bars on the windows within one-half block (roughly 300 feet) of the front of the building in which the apartment is located. This reduced rents by $61 a month for the example in Figure 5. Trash or litter in the neighborhood had a significant negative impact as well. 

The graphs in this article illustrate only a few combinations of home and neighborhood features that the model can analyze. Many other permutations are possible. NAHB multifamily members interested in some of the other possibilities, or who want more details about the statistical model used to generate the gross rent estimates, may contact Paul Emrath, Assistant Staff Vice President of Housing Policy Research, by e-mail, or by calling 800-368-5242, x8449. NAHB also is studying the possibility of making the model available to Multifamily members on an interactive basis through the NAHB web site.

---------------------------------------------------------------------------

¹ The first version of the model was developed in 1993.  Work related to it has been published in several NAHB articles:

^{Paul Emrath, “Quality and Price for New and Existing Homes,” Housing Economics, August 1993.
Paul Emrath, “Features That Influence Home Values,” Housing Economics, December 1995.
Paul Emrath, “Explaining Differences in Market Rents,” Housing Economics, July 1996; reprinted in RAM Digest.
Paul Emrath, “Calculating the Value of Apartment Features and Amenities,” Multifamily Market Outlook, October 2001.
Paul Emrath, “Explaining House Prices,” Housing Economics, January 2002.
Paul Emrath,  “The Need for New Housing Revisited,” Multifamily Market Outlook, March 2002.}

²  The primary technique is called hedonic regression, a term used loosely to refer to a method  that estimates the price of a good based on its characteristics.  Hedonic price estimation dates back at least to Waugh (1928), although Griliches (1961) and Rosen (1974) are usually credited for establishing it as a widely used technique.  Theoretic work on the technique continues, including a recent paper by a Nobel laureate in economics (Heckman et al, 2003). 

References:
Zvi Griliches (1961) “Hedonic Price Indexes for Automobiles: An Econometric Analysis of Quality Changes,” National Bureau of Economic Research, The Price Statistics of the Federal Government, pp. 173-196.

^{James J. Heckman, Rosa Matzkin, and Lars Nesheim (2003) “Simulation and Estimation of
Nonadditive Hedonic Models,”  National Bureau of Economic Research Working Paper 9895.}

^{Sherwin Rosen (1974) “Hedonic Prices and Implicit Market Product Differentiation in Pure Competition,” Journal of Political Economy,  pp. 34-55.}

^{F.W. Waugh (1928) “Quality Factors Influencing Vegetable Prices,” Journal of Farm Economics, pp 185-196.}

 ³ Some builders have reported that elevators are almost ubiquitous in buildings with more than three stories, but the AHS data on which the estimates are based contain many examples of four-to-nine story buildings with tenants who report that they lack access to a working elevator. 

⁴ Paul Emrath, “Home Vintage and Operating Costs,” Housing Economics, November 1997.

⁵ NAHB attempts to make the model as useful as possible for these purposes.  A large number of apartment features is investigated, in order to avoid excluding any arbitrarily.   Generally only those that satisfy statistical tests are retained, but many alternative permutations of the model are investigated, even with variables that don’t satisfy those tests, to minimize the extent to which features included in the model may be acting as proxies for those that are excluded.  For a product as complex as a rental apartment, however, it’s never possible to be sure this has been done perfectly.