HousingEconomics.com Resources
New Version of NAHB’s House Price Estimator Released

Special Studies, August 3, 2015
By Paul Emrath, Ph.D.
Economics and Housing Policy
National Association of Home Builders
Report available to the public as a courtesy of HousingEconomics.com

NAHB maintains a statistical model for estimating the price of a single-family detached home based on its characteristics. Since 2004, NAHB has made the model available online, as an interactive tool viewers can use to compare the cost of different amenity packages, show how characteristics of potential building sites affect the ultimate sales price of a home, and see how polices to clean up neighborhoods are likely to affect property values.

The estimator is calibrated with data from the American Housing Survey (AHS, a nationally representative survey of housing units funded by the Department of Housing and Urban Development and conducted by the U.S. Census Bureau) and usually updated when a new version of the AHS becomes available (every other year). The latest version of the house price estimator has just been released, based on data from the most recent 2013 AHS.

This article describes the new estimator and illustrates some of the results that can be generated from it for a standard recently built (in 2010 or later) home with typical characteristics. Highlights include

  • The price of the standard recently built home varies from $142,000 if built outside of a metropolitan area in the Midwest Census region to $444,800 in a suburb of one of the large California metro areas.
  • If the standard home is built in a Southern central city, adding an extra full bathroom to it will boost its value from $175,500 to $210,800—an increase of about $37,000.
  • Locating the home on a body of water will boost its value to $238,400—an increase of nearly $65,000.
  • Several neighborhood effects are particularly strong within in central cities of metropolitan areas. This may be in part a function of certain neighborhood features being available mostly in larger, higher-priced cities.
  • For example, being located near a subway or other rail system will boost the value of the standard home by $30,000 in a southern suburb, but by more than $71,000 in a southern central city. However, homes near commuter rail tend to be concentrated in particular parts of the country—over 60 percent are in the Chicago, Boston, Philadelphia, New York City-Northern New Jersey, and San Francisco-Oakland Chicago areas—and this effect may by partially acting as a proxy for other factors that make housing expensive in areas like these.

Subsequent sections describe these and other results in more detail, after providing basic background on the model underlying the House Price Estimator.

Estimator Details

The House Price Estimator is a user-friendly version of a statistical model developed and maintained by NAHB to control for the impact various characteristics have on the value of a home[1]. The model is based on values of existing homes, so the estimated price reflects demand for particular features. In a market equilibrium, this should be driven to the cost of installing the features in an existing home. But if a feature has just been introduced into the market or has become increasingly popular recently, it may take a while for enough remodeling to occur to restore equilibrium, and the estimated value may temporarily diverge from its cost.

As in many federal surveys, the value of a home in the AHS is based on owners’ estimates. Several studies have investigated how accurate owners actually are in practice[2]. A typical finding is that owners’ estimates are often off by a few percentage points, but the error is not related to characteristics of the home or neighborhood, allowing the relative impacts of various features on price to be estimated with reasonable accuracy.

A notable strength of the AHS is the amount of detail it contains on each home, its occupants, and neighborhood all combined in a single data set. Indeed there are nearly 3,000 different variables in the 2013 public use AHS file.

Even so, no conceivable data set or model can capture every relevant feature for a commodity as large and complex as a house, so there's always a chance that some particular feature in the model is acting partly as a proxy for others. In technical literature, this is called omitted variable bias, and is another reason the estimated value of a feature may differ from its cost.

NAHB’ approach to estimating house price impacts is to experiment with many combinations of AHS home and neighborhood characteristics, and the interactions among them, retaining those with a statistically significant and plausible impact on price, without excluding anything that would change other estimated impacts in a meaningful way. The goal is to exploit the wealth of information available in the AHS while reducing omitted variable bias as much as possible. For researchers interested in the actual estimated coefficients and their statistical significance, a technical Appendix is available in the “Additional Resources” box.

Region and Metro Status

Figure 1 illustrates the more user-friendly online version of the model, using the features and estimated price for a standard recently built (in 2010 or later) home in a southern central city as an example. The South is chosen as the baseline region for reference, because it accounts for roughly half of housing starts. A central city location is chosen; because, as later sections show, a number of neighborhood effects are particularly strong in central cities.

Figure 1. Characteristics and Estimated Price of the Standard Home

Click image for larger view

The estimated price of $173,530 for the standard recently built home in a southern central city will serve as a basis of comparison for homes with other features or in different locations.

NAHB’s house price estimator uses whatever geographic information is available in the AHS. The AHS identifies if a home is in one of the four principal census regions, but not specific states. Some of the larger metropolitan areas (collections of counties defined by the U.S. Office of Management and Budget based on commuting patterns) are identified, but there are generally too few observations in any one metro to treat separately. However, NAHB has been able carve out a number of large metro areas in California[3] and treat them as a separate “region."

Even when it doesn’t name the specific metro, the AHS does identify if a home is in some metropolitan area; and, if so, whether it’s in the metro’s central city[4]. NAHB defines the area inside a metro but outside of its central city as suburban. This is a fairly conventional practice (although the federal government doesn’t have an official definition of suburb).

Figure 2 shows how the estimated price of the standard recently built home varies across the 14 region/metro status combinations. House price tends to be higher in the Northeast and West Census regions, as well as inside metro areas, with particularly high prices in the large California metro areas (demonstrating the utility of treating them as a distinct “region”).

Figure 2. Estimated Price of a Standard Recently Built Home by Location

Click image for larger view

The standard recently built home costs only about $142,000 if built outside of a metropolitan area in the Midwest Census region, but over $400,000 if built in a central city or suburb of the large California metros. In a given region, the price is higher in suburbs than in central cities, although in some cases the difference is relatively small. In the Midwest, the difference between the standard home in a suburb and central city is less than $3,000 or about 2 percent (compared to $46,000/12 percent in the California metros).

It’s important to remember that the price generated by the estimator is an average across a Census region, rather than the price of a specific home in a specific neighborhood. House prices within a region vary. Prices of similar homes in Oklahoma City and Washington D.C., for instance, may be quite different, even though both locations are central cities in the South Census region.

Physical Features of the Home

Figure 3 shows how the estimated price of a standard recently built home in a southern central city changes with the home’s physical features. Holding square footage and other features constant, a bed, dining or miscellaneous room changes the estimated price by $10,000 or slightly less.

Figure 3. Price of Recently Built Home in a Southern Central City

Click image for larger view

The impact on price is about $17,000 for a family room and $19,000 for a half bathroom. A half bath has hot and cold running water, and either a toilet or bathtub/shower, but not both. If it has both a toilet and a bathtub or shower, the Census Bureau classifies it as a full bathroom. Of the features shown in Figure 4, an extra full bathroom has the largest impact, increasing the estimated price of the standard home by about $37,000—very close to twice the impact of a half bath.

Price of Recently Built Home in a Southern Central City: Impact of Changing Neighborhood Features

Click image for larger view

The house price estimator captures an interesting interaction between bedrooms and bathrooms, indicating that the preferred configuration for a home is one more bedroom than bathroom. Owners appear to attach somewhat less value to additional rooms as the home diverges from this ideal. In Figure 3, adding a fourth bedroom to a standard home with 2 bathrooms increased its value by 5.6 percent. If the standard home instead started with only 1 bathroom, adding a fourth bedroom would increase its value by 4.3 percent. All else equal, the third bathroom makes the fourth bedroom more valuable (and vice versa).

Other than an extra full bath, the largest price impacts in Figure 3 come from adding a full basement (increasing the value of the home by $34,000) and fireplace ($27,000) to the standard home.

It’s important to remember potential (omitted variable) effects of characteristics not captured in the model. For example, if homes with fireplaces were more likely to have tray ceilings, crown molding or other decorative trim (features not available in the AHS) the estimated value of the fireplace could in part be picking up the value of these omitted decorative features.

Location, Location, Location Revisited

NAHB models have repeatedly demonstrated how important neighborhood environment is when estimating the value of a home. A major strength of the AHS over the years has been the number and nature of neighborhood characteristics it captures. Unfortunately (and over persistent objections from NAHB), HUD and the Census Bureau stopped collecting neighborhood information in the 2011 AHS. However, NAHB continued to lobby, and in 2013 the AHS reinstated many of the old neighborhood characteristics, as well as introducing a few new ones. Figure 4 shows most of those that have a significant effect on house price, again using a recently built home built in a southern central city as the baseline reference case.

As in previous versions of the NAHB estimator, a location on the waterfront and abandoned buildings in the neighborhood have a strong effect on house price—in opposite directions. Moving the standard home to an otherwise similar neighborhood but on the waterfront increases its value by nearly $65,000. Abandoned buildings within a half block reduce its value by more than $33,000.

Three of the new neighborhood characteristics introduced in 2013 also turned out to have a significant effect on house prices. Personal services (such as a hair or nail care salon or drycleaner) accessible by walking increase the standard home’s value by over $10,000. If entertainment (going out to eat, attending a cultural or sporting event, etc.) is accessible by walking, the value goes up by nearly $20,000.

Although previous incarnations of the AHS included a general question on public transportation, the 2013 survey introduced an entirely new set of questions, allowing NAHB to generate a more refined (and stronger) estimate. The new estimator shows that a subway, elevated train, street car, light rail, trolley, commuter or inter-city train within a mile increases the estimated value of the standard home by $71,000—more than any other feature in Figure 4.

This is another case where it’s important to remember that one feature can be acting partly as a proxy for others in the model, especially with a feature like commuter rail, which is not available uniformly across the country. There are well over 300 metropolitan areas in the United States, and not all of them have commuter rail systems. Over 60 percent of the homes within a mile of a rail system are concentrated in a handful of areas (Chicago, Boston, Philadelphia, New York City-Northern New Jersey, and San Francisco-Oakland). Hence, proximity to rail may be picking up other attributes that make housing more expensive in cities like these.

Several of the neighborhood features in the AHS have a particularly strong effect in central cities. In fact, a park or other open space, and entertainment within walking distance only have a detectable effect on house prices if the homes are located in a central city.

In two cases, a central-city neighborhood effect is present in suburbs, but to a lesser degree. If the standard home were in a southern suburb rather than central city, a subway within a mile would increase its value by $30,000 (vs. over $71,000 in a southern central city). Unsatisfactory police protection would reduce its value by $5,000 (vs. more than $15,000), even though, as Figure 2 showed, the starting base price of standard home in a southern suburb is about $5,000 higher. Outside of the central cities and suburbs of metropolitan areas, neither subways nor police protection have a measurable effect on house prices.

Many of the neighborhood effects in NAHB’s house price estimator can be useful in a local public policy context. The policy applications of the negative impact unsatisfactory police protection has on home values, especially in central cities, are fairly obvious, in view of recent well-publicized incidents involving the police in several U.S. central cities.

The way home values rise with proximity to subways and other commuter rail systems also has policy implications, although this can be a two-edged sword. On the one hand, it helps establish the benefits of locating rail near homes and vice versa. On the other, it shows the difficulty of providing affordable housing near these transit nodes.

Using the Estimator Online

This article can only illustrate a few of the possible applications of NAHB’s house price estimator. Adventuresome readers who would like to experiment further are encouraged to try the House Price Estimator for themselves on NAHB’s web site.

The online estimator is a “macro-enabled” Excel workbook (i.e., a file with an .xlsm extension). To run the estimator, you need to have a recent enough version of Excel—with security adjusted to allow macros to run before you try to open the estimator online. Be sure to start on the landing page and read the instructions at the bottom before clicking on the link that opens the estimator.

[1]The term hedonic regression is usually applied to models that estimate the price of a good such as a house based on its characteristics, although hedonic regression has a stronger footing in economic theory than this informal usage implies. Early versions date back at least to Waugh (1928), but theoretical underpinnings provided by Griliches (1961) and Rosen (1974) are usually credited for establishing hedonic regression as a widely used technique.
[2]See for example Kiel and Zabel (1999) The Accuracy of Owner-Provided House Values.
[3]The metro areas included in the California region are Bakersfield, Fresno, Los Angeles-Long Beach, Modesto, Oakland, Orange County, Riverside-San Bernardino, Sacramento, San Diego, San Francisco, San Jose, Santa Barbara-Santa Maria, Santa Rosa, Stockton-Lodi, Vallejo- Fairfield-Napa, and Ventura.
[4]A central city is the primary or largest city within the market, or metropolitan statistical area. The geographic area covered is the entire city limits. In 2003, the Office of Management and Budget officially retired the term and replaced it with “principal place,” a similar but not identical concept. Technical criteria imposed by the Census Bureau to protect respondent confidentiality have so far prevented the AHS from following suit. The term central city is still in fairly common use among city planners.

For more information about this item, please contact Paul Emrath at 800-368-5242 x8449 or via email at pemrath@nahb.org.

Recommend This: Recommend This