The academic ﬁeld of property valuation research crosses many disciplines, and includes not only economics, geography, urban planning and design, but also business, ﬁnance, statistics and even specialized real estate departments and schools. As a result, research in the ﬁeld is often moving in many directions at a single time, as some innovations are following those of the associated disciplines mentioned above while others emanate from the large commercial sector that deals with real estate on a daily basis. Given the dynamic nature of this ﬁeld, periodic reﬂections on current developments can offer great insight into the world of academic real estate valuation research.
We have identiﬁed three separate but related trends in valuation. The ﬁrst, and perhaps most prevalent, is the increased use of advanced spatial methods in published studies. Second, we see the recent interest in various land value issues, including the focus on land values as a major shaper of both real estate values and of urban spaces in general, as another major trend. The third is the measurement of value premiums offered by energy efﬁcient, sustainable, or green locations and buildings. We address each trend below, followed by a concluding synthesis.
Spatial methods in valuation
The importance of space or location in determining real estate values is, to even the novice, axiomatic. Properly incorporating space into valuation models, however, is not without its difﬁculties. Spatial dependence, spatial heterogeneity, anisotropic phenomena and boundary effects all combine to render obsolete the simple non-linear decline of values from the central business district (CBD), as assumed in the basic monocentric urban economic model. In place of this stylized circular city are polycentric urban regions complete with localized amenities (or disamenities), geographic heterogeneities, fragmented municipal governments, and complex systems of land use regulations.
To deal with these inherent complexities, a host of advanced spatial methods have been developed over the last few decades. These tools can roughly be divided into those dealing with spatial dependence and those dealing with spatial heterogeneity. The number of papers on these topics is vast. We cover a select few here to illustrate the breadth of uses and variations on them that are occurring in the academic research.
There is a long history of attempts to account for spatial effects in real estate valuation studies. Early work by Goodman (1978) and Li and Brown (1980) examined spatial effects at the scale of the neighborhood. Building on this, Dubin (1988), and bolugbe (1997), among others, presented work that considered property-speciﬁc impacts of spatial dependence in the market. Throughout the late 1980s and early 1990s, more advanced spatial econometrics methods were being developed in other disciplines (see Anselin, 1988) which eventually made their way into the ﬁeld of property valuation research. Real estate, an industry whose deﬁning mantra is considered ‘‘location, location, location,’’ is well suited to beneﬁt from these advances in spatial econometrics (hen & Coughlin, 2008).
Two basic models underlie attempts to model spatial dependence, the spatial lag (or spatial autoregressive) model and the spatial error model. Spatial lag models allow for the observed value of nearby observations—in the case of real estate, nearby sales prices— to impact the dependent variable in the models. These lag models attempt to capture the spatial dependence in the real estate market, or in other words, account for the impact of nearby sales on current home prices. Given the fact that asking prices for homes on the market are often the direct product of nearby home prices (Brasington & Hite, 2005) and that appraisers determine values for ﬁnancing purposes based on nearby comparable sales (cia, & Piras, in press) it seems natural that spatial dependence would play a role in home price estimation. Spatial error models (SEMs), on the other hand, deal speciﬁcally with spatial autocorrelation in the error terms, a violation of the basic assumptions of ordinary least squares (OLS) regression. By accounting for the spatial autocorrelation in the disturbances, SEM models can help to eliminate omitted variable bias generated by missing spatial variables.
Spatial econometrics is not limited to the two basic models. By combining spatial lag and spatial error models, a general spatial model (GSM) can be estimated. If spatial dependence is suspected to exist in both the dependent variable and the explanatory variable(s) a Spatial Durbin model is the most appropriate speciﬁcation (LeSage & Pace, 2009).
Though spatial econometrics has developed into a complex, multi-faceted analytical technique, it does not need to remain unavailable to the general practitioner of valuation-related research. Recent work by Osland (2010) provides a useful overview of spatial econometric models with a focus on real estate valuation. Using house price data from Norway, Osland works through the speciﬁcation of various spatial models as well as a discussion on application and interpretation of a set of diagnostic tests aimed at choosing the proper spatial speciﬁcation. Her comparative analysis shows that these spatially explicit models vastly outperform a traditional OLS model.
Where Osland’s work is primarily concerned with the predictive accuracy of house prices, other recent research uses spatial econometrics in a valuation framework to isolate the price impacts of various variables of interest. Brasington and Haurin (2006) ﬁnd the impact of school spending on home prices is larger when using a spatial lag model that accounts for spatial dependence than with a basic OLS model including neighborhood descriptors. Cohen and Coughlin (2008) compare spatial lag, spatial error and general spatial models in a study on the impacts of airport noise on home prices in Atlanta, Georgia. They ﬁnd that the coefﬁcient estimates of impact from airport noise do not differ signiﬁcantly in the spatial models from a baseline OLS model, but that when the spatial multiplier effect is considered the traditional models are shown to be likely underestimating the impact of airport noise on housing prices. Kuethe (2012) uses spatial lag and spatial error models to examine the impact of land use diversity and land use fragmentation on house prices in Milwaukee, Wisconsin. His baseline OLS model suggests that land use diversity has a positive impact on house prices whereas the impact of land use fragmentation shows no statistically signiﬁcant relationship. After accounting for spatial autocorrelation in the error terms with a spatial error model, Kuethe’s updated model shows that land use diversity has no effect and high levels of fragmentation are related to increases in housing prices. In a study on the magnitude of welfare effects as related to housing attributes (including location), Koschinsky et al. (in press) ﬁnd that estimates from a basic OLS model, and OLS models with ﬁxed spatial effects, are different enough from those found with spatial econometric models to be misleading to the researcher.
Their work suggests that the spatial dependence between observations (in this case home prices) is not properly accounted for by ﬁxed spatial effects such as discrete geographic boundaries.
These studies exhibit the dangers that can result from ignoring spatial effects and proceeding with basic OLS models when valuing real estate. In each case, adding in spatial effects through spatial lag or spatial error models signiﬁcantly changed the magnitude of the effect or even the ﬁnding of an effect at all. This work–and many other similar projects—indicates that in research on real estate pricing the impacts of spatial dependence should not be ignored.
Over the past decade spatial econometrics has become ﬁrmly entrenched in the real estate valuation literature. However, a nagging question has plagued the ﬁeld for years: How is the proper spatial weights matrix determined? In a recent paper, LeSage and Pace (2011) show that, in contrast to published work, in most situations the speciﬁcation of the weight matrix does not substantially change the estimates of the model. They have deemed this pervasive idea— that ﬁnding the ideal weight matrix is critical—as the ‘‘Biggest Myth in Spatial Econometrics.’’ Their work shows that two pitfalls often lead to this erroneous conclusion. First, because coefﬁcient estimates should not differ between a correctly speciﬁed OLS and SEM model (differences should show up in the measures of error), large changes in coefﬁcient estimates due to changes in weight matrices are a sign of model misspeciﬁcation rather than a response to matrix speciﬁcation. Second, the incorrect interpretation of b coefﬁcients in spatial lag models will lead researchers to notice a large change in effects due to small variations in the spatial weight matrix. As LeSage and Pace show, a properly speciﬁed model containing spatial lags will use variations in the b and q estimates to keep the true effect estimates (partial derivatives) relatively constant over changing weight matrices. In sum, the authors do not suggest that the identiﬁcation of weight matrices makes no difference at all, but rather that the two common errors discussed above are more likely to blame for large variations in effects estimates than the particular choice of the spatial weight matrix.
The spatial econometric models discussed above have emerged to deal with the spatial dependence of nearby observations. Another set of modeling techniques seek to examine if the relationships between independent and dependent variables vary across space. Such models have a long history, dating back to spatial expansion methods developed by Casetti (1972) and later used by Can (1990), and a related variable interaction approach presented by Fik, Ling, and Mulligan (2003). Recently, the most common technique to deal with spatial heterogeneity in the coefﬁcient estimates has been the use of local regression models (LOESS). Geographically weighted regression (GWR), as elaborated by Fotheringham, Brunsdon, and Charlton (2002), is the most prevalent of these methods. Since the early 1990s, local regression models have been used widely, examples being work by McMillen (1996), Pavlov (2000), and Bitter, Mulligan and Dall’erba (2007). Hannonen (2008) describes these local techniques as being ‘‘data-driven and ﬂexible’’ and notes that their employment can allow the user to limit concerns over the choice of functional form. Hannonen’s (2008) recent work on forecasting land prices in Espoo, Finland ﬁnds that a robust form of local regression model outperforms a traditional OLS model, but not a structural time series model. Paez, Long, and Farber (2008), working with home sales data from Toronto, also ﬁnd that localized models (GWR and a moving window regression, MWR) produce better out-of-sample home price predictions than a base OLS model and a moving window Kriging method. Conversely, Osland (2010) ﬁnds that while her GWR model does show evidence of heterogeneity in the coefﬁcients over space, the predictive results do not differ signsﬁcantly from the base OLS model and are outperformed by models accounting directly for spatial dependence.
Not all local regression models are used speciﬁcally for price prediction purposes. Sunding and Swoboda (2010) use a GWR to model the spatial heterogeneity in housing ‘‘shadow prices’’ that result from land use regulations in the Los Angeles metro region. Their ﬁndings suggest that failure to account for spatial heterogeneity in the model can lead to an under-estimate of the effect of land use regulations on home prices, especially for high end homes. Cho, Lambert, Roberts, and Kim (2010) use both a GWR model and a GWR model combined with a spatial error model (SEM) speciﬁcation to estimate the marginal rate of substitution of shared open space for parcel size in Knox County, TN. They ﬁnd that the GWR-based models produce more accurate results than the baseline OLS speciﬁcation and that coefﬁcients for the marginal rate of substitution are signiﬁcant in the GWR speciﬁcation but not in the OLS model. Finally, Du and Mulley (2006) use a GWR-based method to examine the spatial heterogeneity of the relationship between urban land values and access to public transportation, using data from England.
Most of the empirical work using local regression models, such as a GWR, shows improvements over baseline OLS models. Further, the theory behind them – accounting for the spatial heterogeneity of regression coefﬁcients—makes intuitive sense from a real estate perspective. These models, however, are not without some concerns. Wheeler and Tiefelsdorf (2005) sound an alarm regarding the multicollinearity of local parameters produced by models such as GWR, and present some preliminary diagnostic tests to measure this. In more recent work, Paez, Farber, and Wheeler (2011) run a number of simulations to test the statistical properties of GWR on datasets with known spatial properties. Their work shows that GWR models have a much higher ‘false positive’ rate of ﬁnding spatial heterogeneity when it actually does not exist and that using GWR on small datasets (less than n = 160) can be problematic. Further, they suggest that researchers should be careful in using methods such as GWR to make inferential determinations about multivariate spatial relationships—especially toward policy ends. In short, GWR and other local regression techniques provide researchers with highly ﬂexible methods to measure the spatial heterogeneity of relationships existing between housing attributes and housing prices; however, more research is needed regarding the pure statistical or inferential qualities of such models.
Emergence of land values as a research interest
The recent housing market boom and subsequent bust in the US have left researchers scrambling to explain the mechanics of the bubble (Glaeser, Gyourko, & Saks, 2005; Wheaton & Nechayev, 2008; McDonald & Stokes, in press). While the causes of the crash are oft-debated, multi-faceted (involving both supply and demand factors) and ultimately beyond the scope of this paper, a number of recent studies have turned to looking more closely at the mechanics of land values in order to better understand the temporal and geographic components of the recent volatility in real estate values. The research in this area can be divided into two broad and related types: studies aimed at tracking land price indices and those looking at decomposing real estate values into land and structural components. Some research investigates both of these themes. We cover the latest research in both areas below, beginning with land value indices.
Land value indices
Home price indices (HPIs) such as S&P/Case-Shiller or Federal Housing Finance Agency (FHFA) are well known throughout the industry and the general media. Similar indices of land value, however, are few and far between. Beginning around 2007, a handful of studies emerged, attempting to expand the popular trend-monitoring approach speciﬁcally to land values. In doing so, Davis and Heathcote (2007) found that real residential land prices increased fourfold from 1975 to 2006, much higher than similar HPIs which showed housing prices doubling over the same time period. Related work by Davis and Palumbo (2008) broke down residential land values by geographic region and found that, in percentage terms, land value increases were largest in the Midwest (208%) from 1984 to 1998, compared to smaller increases experienced in the Southeast, East Coast and West Coast (all around 50%), while the Southwest actually declined over this period (-17%). The more recent time period from 1999 to 2004 shows much larger gains on the coasts—East Coast, 115% and West Coast, 145% – than in the rest of the country (50–75%). Although the exact trends vary between the different studies, this is primarily an artifact of the varying methods used to extract and monitor land price changes. Regardless of the speciﬁc results, the overall message is clear: from the 1970 to the mid-2000s, residential land values increased much faster than overall home prices and it is certain that these price trends have a geographic component.
More recent work has shed light on how these trends have fared in the aftermath of the housing bubble and if the same trends carry over into other land use categories. Using disaggregated data from 23 US metropolitan areas, Oliner, Nichols, and Mulhall (2010) ﬁnd that residential land prices showed greater price increases prepeak, and greater price decreases post-peak, along the coasts— those areas with higher overall land prices. Sirmans and Slade (in press) show similar price movement trends, including the ﬁnding that land price indices tend to ‘lead’ housing price indices, meaning that during the latest bubble, land prices peaked prior to home prices.
Similar, though less pronounced, trends are found for commercial and industrial land. Sirmans and Slade (in press) show that, using 1990 as a base (equal to 1.0), their transaction-based index of commercial land values peaked in late 2006 at just over 2.5 and industrial land topped out at around 4.0 at about the same time. These compare with a peak of over 4.5 for residential land in mid-2005. Oliner et al. (2010) combine commercial and industrial prices and also show that these non-residential land values both peaked later (early 2007 versus early 2006) and lower (111% versus 166%) when compared to residential land values.
This work indicates that land values have experienced tremendous movement over the past four decades and that these movements dwarf those undergone by improved properties in the real estate market. Some of these studies, and a number of subsequent follow-ups, take a deeper look at land values and how to separate them from improvement values, or how to decompose real estate values into its various parts.
Real estate value decomposition
The simple decomposition at the heart of this work involves dividing real estate values into a land component and an improvement component. Some of the earliest proponents of this method— Davis and Heathcote (2007) and Bostic, Longhofer, and Redfearn (2007)—cite the important theoretical observation that different factors inﬂuence the two components and, therefore, we should expect that land values and improvement values will respond differently to external shocks. Further, Davis and Heathcote (2007) explain that improvement values are nothing more than depreciated estimates of construction costs. As such these values are tethered to a general index of material and labor costs and since materials and labor compete in the larger market for other goods, and are generally mobile, as a result construction costs (improvement values) should generally mimic larger economic trends in terms of price movements (Oliner et al., 2010). Land, on the other hand, is spatially ﬁxed and locally supply-constrained, which means that land values are much more susceptible to large volatility arising from external demand shocks.
From this underlying theory, Davis and Heathcote (2007) and Bostic et al. (2007) develop a similar central hypothesis: residential properties with higher ratios of land values to improved values will show more price volatility. Subsequent research has expanded on this simple hypothesis to discuss the effects of land leverage conditional on time, geographic region, property age, and property type. The results of these studies are discussed below.
Price volatility and land leverage
The original land leverage research (Bostic et al., 2007; Davis & Heathcote, 2007) and the handful of subsequent studies that we have reviewed all validate the basic hypothesis that higher land leverage is consistent with higher price volatility. In looking at aggregated national-level data from 1975 to 2006, Davis and Heathcote (2007) show that residential land values have been 2.2 times as volatile as overall housing prices and 3.3 times as volatile as improved values. Given that changes in home prices are a weighted average of changes in land and improvement prices, properties with higher land leverage will therefore be more price volatile than those with lower land leverage. By disaggregating their analysis into the 46 largest US metropolitan areas, Davis and Palumbo (2008) show that metropolitan-wide residential price volatility is related to metropolitan-wide residential land leverage, suggesting that increases in land values were the driver behind the intense increases in home prices from 1998 to 2004 (and beyond). Similarly, a more recent study by Oliner et al. (2010) suggests that the relationship between land leverage and real estate prices continues to hold during signiﬁcant down-cycles in the US market (2007–2009) as well. Bourassa, Hoesli, Scognamiglio, and Zhang (2011) show that this the tie between land leverage and home price volatility is not solely an American phenomenon. Their research using Swiss national level data over the period from 1978 to 2008 also links land leverage to home price volatility in both boom and bust periods.
Another ﬁnding to come out of this literature is that land leverage has been steadily increasing over time (the current US real estate bust excluded). Davis and Heathcote ﬁnd that, nationally, residential land leverage rose from an average of 10.4% in 1950 to 36.4% in the year 2000. Looking at land leverage trends by geographic region, Davis and Palumbo show signiﬁcant increases in residential land leverage from 1984 to 2004 for the Midwest (11–36%), the Southeast (27–42%), the East Coast (38–64%), and the West Coast (55–74%). Only the Southwest showed a diminutive gain, increasing from 35% in 1984 to 38% in 2004. Again these trends are not limited to the United States. A study of Helsinki, Finland shows that average residential land leverage in the metropolitan region increased from around 40% in 1988 to greater than 50% in 2007 (Oikarinen, 2010). Likewise, Bourassa et al. (2011) show an average growth throughout all of Switzerland in residential land leverage from 32% in 1978 to 53% in 2007. Given the relationship between volatility and land leverage, these ﬁndings suggest that the recent price volatility experienced in markets around the world is here to stay, so long as land leverage rates remain high (Davis & Palumbo, 2008).
Due to the existence of systematic variations in land values over space (price gradients of some shape) and therefore the differences in the substitutability of land and capital over space, theory suggests that land leverage should also vary systematically over space (Bostic et al., 2007). The depreciation of older structures which are generally concentrated in the center of our cities adds to the likelihood of distinct spatial patterns in land leverage. Bostic et al.’s ﬁnding bear this out as they ﬁnd higher land leverages in older, more central neighborhoods than in newer, peripheral suburbs. Other attempts at empirically deﬁning an intra-urban spatial component to land leverage are mysteriously lacking and represent a potential avenue of new research.
Sustainability and property values
During the past decade sustainability has emerged as a dominant topic of interest among urban scholars. A voluminous body of literature has sprung up to demonstrate the problems associated with urban sprawl and the beneﬁts of energy-efﬁcient buildings and more compact and connected forms of development as embodied in New Urbanism, traditional neighborhood development (TND) and transit-oriented development (TOD). The superiority of these forms of development is now nearly universally accepted by academics and urban planners – but they have yet to make substantial inroads into the marketplace. This may be due in part to the fact that they are more difﬁcult and costly to build – and will not be provided by the private sector unless a clear economic incentive exists. In response, a small but growing body of literature has sought to demonstrate that green buildings and compact development can command pricing premiums over more conventional forms of development.
Green building premiums
It is widely recognized that urban buildings are a major contributor to energy consumption and greenhouse gas emissions. Moreover, substantial resources and energy are embodied during their construction. However, buildings are not all created equal – some are more efﬁcient in their use of resources than others. In the United States, two voluntary certiﬁcation programs have been developed to help building consumers differentiate ‘‘green’’ buildings— Energy Star and LEED. Energy Star ratings are based on energy efﬁciency, whereas LEED designations include many other sustainable design attributes in addition to operating energy efﬁciency. A number of potential economic beneﬁts associated with green-certiﬁed buildings may enhance values – of which energy cost savings are the most obvious. Others include higher labor productivity for green-building workers and marketing advantages to occupiers in the form of enhanced corporate image. To the extent that these beneﬁts exist, green building occupiers should be willing to pay higher rents, and because commercial buildings are valued based on the income stream they generate, rent premiums should be capitalized into building values. Moreover, green buildings could command pricing premiums if they are perceived to be less risky. For example, there may be lower risk of incurring compliance costs in the event that energy efﬁciency regulations are enacted in the future.
Three recent studies, Wiley, Beneﬁeld, and Johnson (2010), Eichholtz, Kok, and Quigley (2010) and Fuerst and McAllister (2011) have sought to quantify the price premiums associated with ofﬁce buildings that have received green certiﬁcations. Their results suggest that green buildings do indeed generate statistically signiﬁcant sales price premiums. The Wiley et al. (2010) study pegs the premium at $30/sf for Energy Star rated buildings and $129/sf for LEED, while Eichholtz et al. (2010) ﬁnd a 16–17% premium for the two types combined. Fuerst and McAllister record a much larger price premium for green buildings of nearly 30%, with relatively little difference between LEED and Energy Star buildings. These premiums are found to stem in part from the greater income streams that green buildings generate – as they enjoy both higher rents and occupancy rates. However, the price premiums are generally greater in magnitude than the income premiums, which suggests that investors perceive beneﬁts from green building ownership above and beyond their ability to generate higher operating income.
The deviations in value premiums reported in these studies (despite their reliance on identical property data) may be partially attributable to variable and/or inadequate controls for location in their pricing models. It is also likely that the premium for green buildings will vary with geographic context. Only the Eichholtz et al. (2010) study considers this possibility by deﬁning ‘‘clusters’’ of buildings within a 1 4 mile radius of each green-designated building and allowing premiums to vary by cluster. While the paper does not report the full results of this analysis, the authors indicate that buildings in smaller markets and peripheral locations of large metropolitan areas generate greater pricing premiums than those located in ‘‘prime’’ areas.
The value of compact development
In addition to growing interest in green building values, urban researchers are increasingly focusing their attention on identifying the pricing premiums attributable to more accessible, connected, and compact urban designs that facilitate walking and reduce the need to drive, as even a net zero energy building is not sustainable if located far from jobs, employees, services and transit. In this section, we highlight a collection of recent studies concerned with this topic.
Urban economists have long been concerned with the roles that accessibility and locational attributes play in shaping land values, but it is only recently that property valuation studies have been cast within the context of sustainability. The voluminous body of research focusing on proximity to ﬁxed rail stations and TOD (see Bartholomew & Ewing, 2011 for a recent review) has demonstrated that the enhanced accessibility afforded by these locations generally increases property values in both the US (e.g. Snow & Kahn, 2000; Cervero & Duncan, 2002; Hess & Almeida, 2007) and abroad (Cervero & Murakami, 2009; Debrezion, Pels, & Rietveld, 2011). However, in some cases negative externalities associated with the station itself outweigh the accessibility beneﬁts (Bowes & Ihlanfeldt, 2001; Goetz, Ko, Hagar, Ton, & Matson, 2010) and premiums for walk-and-ride stations are generally found to be greater than for park-and-ride (Bollinger & Ihlanfeldt, 1997; Gatzlaff & Smith, 1993; Kahn, 2007; Landis & Zhang, 1995). A separate strand of the literature has shown that neighborhood land use mix and proximity to non-residential uses inﬂuence property values (Cao & Cory, 1981; Grether & Mieszkowski, 1980; Li & Brown, 1980; Song & Knaap, 2004; Mathur, 2008). Mixed uses and proximity to retail properties appear to generate property value premiums in some cases but, again, the accessibility beneﬁt may be offset by negative spillover effects emanating from nonresidential properties (Li & Brown, 1980). Other researchers have focused on the role of neighborhood design and street layout (Asabere, 1990; Guttery, 2002; Tu & Eppli, 1999; Song & Knaap, 2003; Ryan & Weber, 2007; Song & Quercia, 2008). These studies yield mixed results – as the features associated with New Urbanism and TND are not always found to elicit higher residential property values.
The assembly of literature referenced above generally supports the notion that consumers are willing to pay premiums for features associated with ‘‘compact’’ development, although these premiums are not universal and their magnitude appears to vary with geographic context. The ﬁrst three recent studies that we detail in this section address the interplay between accessibility, land use mix, and neighborhood design. They are important because they demonstrate that synergies may exist between these elements, and represent a starting point for understanding the contexts in which more compact forms of development may achieve market acceptance.
Matthews and Turnbull (2007) examine how street connectivity, land-use mix, and proximity to retail land use inﬂuence single-family home values in King County, Washington. Their hedonic models demonstrate that neighborhood design features interact with one another in the determination of housing prices. A more grid-iron like street pattern generates a price premium on the west side of the study area, which includes Seattle and its more pedestrian-oriented neighborhoods. Within this area, the premium associated with proximity to retailing increases when combined with greater street connectivity, though it varies by the speciﬁc measure of connectivity used. In contrast, both gridlike neighborhood street patterns and proximity to retailing reduce housing prices in the eastern portion of the study area, which is composed of auto-oriented suburbs. The authors conclude that features associated with New Urbanism do not have universal appeal and might not make sense in all areas.
Recent papers by Atkinson-Palombo (2010) and Duncan (2011) explore how neighborhood design interacts with proximity to light rail transit (LRT) stations in the determination of housing values. They demonstrate that the premium attributable to station proximity is enhanced when the station is located within mixed-use pedestrian-oriented environments. The Atkinson-Palombo study focuses on ‘‘pre-service’’ price impacts of the new Phoenix LRT and shows that premiums for proximity to a station are conditional upon both the type of property and neighborhood context. She ﬁnds that single-family homes within walking distance (deﬁned as 1 4 mile) of an announced walk-and-ride LRT station command premiums of about 6% in ‘‘amenity-dominated mixed-use neighborhoods’’ – while homes near stations (typically park-and-ride) in predominantly residential neighborhoods command no premium, and actually sell at a discount if they are also located in a designated TOD overlay zone. Condos within walking distance of an LRT station in mixed-use neighborhoods achieve much greater premiums (16%) than homes, and those also located in overlay zones command an additional 37% premium. Conversely, condos in residential neighborhoods sell at only a slight premium. Atkinson-Palombo argues that the results at least in part reﬂect locational sorting – as residential neighborhoods may not be attractive to consumers seeking proximity to LRT. Duncan (2011) ﬁnds largely supportive results in a study focused on service-stage impacts of station proximity in a hedonic analysis of condo prices near LRT stations in San Diego. He demonstrates that station proximity premiums are conditioned by the quality of the pedestrian environment in which the unit is located, which is measured by four separate variables: (1) density of street intersections; (2) density of population serving employment; (3) steepness of terrain; and (4) area dedicated to park-and-ride lots. Of these, the presence of population serving employment appears to have the greatest synergistic effect with station proximity. The magnitude of the interaction between pedestrian orientation and station proximity is assessed by comparing a hypothetical ‘‘good’’ pedestrian neighborhood, with 75th percentile values on all four measures, to a ‘‘bad’’ neighborhood with 25th percentile values. Condos located 0.3 km from a walk-and-ride LRT station sold at an estimated premium of $20,000 over those 1.6 km from a station if they were located in a neighborhood deemed to have a good pedestrian environment, while those in a poor pedestrian environments were discounted by nearly $10,000, indicating that LRT is perceived to be a disamenity within this context. The comparable premium is somewhat smaller, and discount larger, for condos situated near a park-and-ride station.
The ﬁrst three studies highlight the synergistic effects associated with combinations of land use, urban design, and access elements – and suggest that consumers are willing to pay premiums for speciﬁc combinations. The ﬁnal two studies we highlight, Rauterkus and Miller (2011) and Pivo and Fisher (2011), focus directly on quantifying price premiums attributable to ‘‘walkability,’’ as opposed to the design and land use attributes that contribute to good pedestrian environments. Both use the now ubiquitous Walk Score to measure degree of walkability. Walk Score rates locations on a scale of 0–100 based on their proximity to various non-residential destinations including schools, retail, food, recreation, and entertainment. Both studies observe that Walk Score is far from a perfect measure of walkability: among other limitations, it only considers straight line distance to the nearest destination of each type, irrespective of its importance or quality, and does not consider walking travel times or the quality of the pedestrian experience.
Both studies show that locations with higher Walk Scores command statistically signiﬁcant property price premiums. In a study of Jefferson County, Alabama, Rauterkus and Miller (2011) ﬁnd that higher Walk Scores are correlated with higher assessed values for residential land, although the study does not draw a clear conclusion regarding the magnitude of this premium. The premiums are also found to be greater in neighborhoods with higher overall Walk Scores than in their less walkable counterparts. Pivo and Fisher (2011) demonstrate that commercial properties with high Walk Scores also command pricing premiums based on a sample of 4200 investor-owned properties located throughout the US. The premiums for a 10 point increase in Walk Score range from 1% for apartment properties to 9% for ofﬁce and retail properties, with no observable effect for industrial properties. The premiums are shown to be higher in less walkable cities, all else being equal. This study also indicates that the premiums derive from the higher incomes (i.e. higher rents) that walkable properties generate, but that buyers do not necessarily reap greater ﬁnancial returns because these beneﬁts have already been capitalized into prices at the time of purchase.
While the results of both studies are encouraging, they should be treated with caution, due to the limitations inherent in Walk Score as noted above. Moreover, both suffer from inadequate locational controls. Locations that achieve high Walk Scores also likely share many other similarities that are unrelated to their walkability and, as Pivo and Fisher (2011) point out, the estimated premiums will clearly be due in part to improved access by other means of transportation as well.
Academic research on real estate valuation is highly dynamic and has capitalized on the recent explosion of cheap computing power and public real estate data. In this paper we have highlighted three prominent trends in real estate valuation; (1) the use of complex spatial modeling techniques; (2) the renewed interest in land values; and (3) the growth in research on the value impacts of sustainable forms of development. The thread that ties all three of these trends together is perhaps the oldest mantra in real estate: location matters. Contrary to the theorizing about how technology would change our need for proximity, the spatial attributes of real estate still remain one of the core concerns of the ﬁeld.
From a methodological standpoint, the emergence of spatial econometrics and related spatially-explicit techniques allow valuation researchers to more completely incorporate the impacts of location into pricing models. These models are necessarily complex and therefore it can be difﬁcult, not only to interpret the results but also to properly specify and estimate them. As the literature suggests, there remains confusion among research practitioners on how exactly to best use these models and what the results of out-
put from the specialized software programs really means for their research question(s). We see the recent articles by Osland (2010) and LeSage (2011) as necessary steps toward increasing the general ease of use and interpretability of these models by nonspecialists.
The recent boom and bust in the US residential (and non-residential) real estate market has researchers searching for causes. One line of research has been the investigation of the differences between movements in land values as opposed to improvement values in the real estate market. This work is based around the fundamental concept that the drivers of land values differ from those of improvements. For urban policymakers, these ﬁndings prove illustrative, since policy decisions most often inﬂuence land values, and land uses, whereas the value and form of the improvements are determined by individuals in the private market. One noticeable omission from this literature, however, is the use of the spatial econometric methods discussed above. Given the dependence of land values on locational attributes, this omission speaks to opportunities here for considerable future research on melding these two lines of inquiry.
The movement toward environmental sustainability is one of the largest, if not the single largest, cross-disciplinary trend of the last few decades. The real estate discipline is no exception, as sustainable real estate, both in location and in building design, has become a sub-industry on its own–one complete with specialized empirical studies, as discussed above. The work to date is encouraging in terms of supporting the economic value of sustainable development; however, these studies suffer from the same limitations as the land value studies—namely the lack of complex spatial methods in the research designs. As the spatial studies show, failure to adequately account for space can create serious bias in coefﬁcient estimates. While this preliminary work is important to setting the ﬁeld of green real estate valuation on its way, the lack of spatial econometric analyses sheds some doubt over the robustness of the results. Additionally, while the green valuation literature generally supports the idea that green buildings and location add a price premium, what is missing is a discussion of the costs of building green. If the costs overwhelm the price premiums in the market, proﬁtability is diminished and private market suppliers will have trouble providing the level of green real estate demanded by the market.
In sum, we ﬁnd these three emerging trends in property valuation research as major steps toward a better understanding of the complex operations of urban real estate markets. However, given the slow incorporation of spatially-explicit methods into these emerging ﬁelds there remain exciting opportunities for researchers looking to make an impact in real estate valuation. The continued expansion of public data availability and quality, along with increased demand for valuation studies, makes this an exciting time to be working in this ﬁeld.
Anselin, L. (1988). Spatial econometrics: Methods and models. Dordrecht: Kluwer Academic.
Asabere, P. K. (1990). The value of a neighborhood street with reference to cul-desac. Journal of Real Estate Finance and Economics, 3, 185–193.
Atkinson-Palombo, C. (2010). Comparing the capitalization beneﬁts of light-rail transit and overlay zoning for single family houses and condos by neighborhood type in metropolitan Phoenix, Arizona. Urban Studies, 47, 2409–2426.
Bartholomew, K., & Ewing, R. (2011). Hedonic price effects of pedestrianand transit-oriented development. Journal of Planning Literature, 26(1), 18–34.
Baum-Snow, N., & Kahn, M. (2000). The effects of new public projects to expand urban rail transit. Journal of Public Economics, 77, 241–263.
Bitter, C., Mulligan, G. F., & Dall’Erba, S. (2007). Incorporating spatial variation in housing attribute prices: A comparison of geographically weighted regression and the spatial expansion method. Journal of Geographical Systems, 9(1), 7–27. Bollinger, C. R., & Ihlanfeldt, K. R. (1997). The impact of rapid rail transit on economic development: The case of Atlanta’s MARTA. Journal of Urban
Economics, 42, 179–204.
Bostic, R. W., Longhofer, S. D., & Redfearn, C. L. (2007). Land leverage: Decomposing home price dynamics. Real Estate Economics, 35(2), 183–208.
Bourassa, S. C., Hoesli, M., Scognamiglio, D., & Zhang, S. (2011). Land leverage and house price. Regional Science and Urban Economics, 41(2), 134–144.
Bowes, D. R., & Ihlanfeldt, K. R. (2001). Identifying the impacts of rail transit stations on residential property values. Journal of Urban Economics, 50, 1–25.
Brasington, D., & Haurin, D. R. (2006). Educational outcomes and house values: A test of the value added approach. Journal of Regional Science, 46(2), 245–268.
Brasington, D., & Hite, D. (2005). Demand for environmental quality: A spatial hedonic analysis. Regional Science and Urban Economics, 35, 57–82.
Can, A. (1990). The measurement of neighborhood dynamics in urban house prices.
Economic Geography, 66(3), 254–272.
Can, A., & Megbolugbe, I. (1997). Spatial dependence and house price index construction. Journal of Real Estate Finance and Economics, 14, 203–222.
Cao, T. V., & Cory, C. D. (1981). Mixed land use, land-use externalities, and residential property values: A reevaluation. Annals of Regional Science, 16, 1–24.
Casetti, E. (1972). Generating models by the expansion method: Applications to geographical research. Geographical Analysis, 4, 81–91.
Cervero, R., & Duncan, M. (2002). Transit’s value-added effects: Light and commuter rail services and commercial land values. Transportation Research Record, 1805, 8–15.
Cervero, R., & Murakami, J. (2009). Rail and property development in Hong Kong: Experiences and extensions. Urban Studies, 46(10), 2019–2043.
Cho, S.-H., Lambert, D. M., Roberts, R., & Kim, S. G. (2010). Moderating urban sprawl: Is there a balance between shared open space and housing parcel size. Journal of Economic Geography, 10(5), 763–783.
Cohen, J. P., & Coughlin, C. C. (2008). Spatial hedonic models of airport noise, proximity, and housing prices. Journal of Regional Science, 48(5), 859–878.
Davis, M. A., & Heathcote, J. (2007). The price and quantity of residential land in the United States. Journal of Monetary Economics, 54, 2595–2620.
Davis, M. A., & Palumbo, M. G. (2008). The price of residential land in large US cities.
Journal of Urban Economics, 63(1), 352–384.
Debrezion, G., Pels, E., & Rietveld, P. (2011). The impact of rail transport on real estate prices: An empirical analysis of the Dutch housing market. Urban Studies, 48(5), 997–1015.
Du, H., & Mulley, C. (2006). Relationship between transport accessibility and land value: Local model approach with geographically weighted regression. Transportation Research Record: Journal of the Transportation Research Board, 1977, 197–205.
Dubin, R. (1988). Estimation of regression coefﬁcients in the presence of spatially autocorrelated error terms. Review of Economics and Statistics, 70, 466–474.
Duncan, M. (2011). The impact of transit-oriented development on housing prices in San Diego, CA. Urban Studies, 48, 101–127.
Eichholtz, P., Kok, N., & Quigley, J. (2010). Doing well by doing good? Green ofﬁce buildings. American Economic Review, 100, 2492–2509.
Fik, T. J., Ling, D. C., & Mulligan, G. F. (2003). Modeling spatial variation in housing prices: A variable interaction approach. Real Estate Economics, 31(4), 623–646. Fotheringham, S., Brunsdon, C., & Charlton, M. (2002). Geographically weighted regression: The analysis of spatially varying relationships. West Sussex: John Wiley
Fuerst, F., & McAllister, P. (2011). Green noise or green value? Measuring the effects of environmental certiﬁcation on ofﬁce values. Real Estate Economics, 39(1), 45–69.
Gatzlaff, D. H., & Smith, M. T. (1993). The impact of Miami Metro rail on the value of residences near station locations. Land Economics, 69, 54–66.
Glaeser, E. L., Gyourko, J., & Saks, R. E. (2005). Why have housing prices gone up?
American Economic Review, 95(2), 329–333.
Goetz, E. G., Ko, K., Hagar, A., Ton, H., & Matson, J. (2010). The Hiawatha line: Impacts on land use and residential housing value (CTS 10-04). Minneapolis, MN: Center for Transportation Studies.
Goodman, A. C. (1978). Hedonic prices, price indices and housing markets. Journal of Urban Economics, 5(4), 471–484.
Grether, D. M., & Mieszkowski, P. (1980). The effect of nonresidential land uses on the prices of adjacent housing: Some estimates of proximity effects. Journal of Urban Economics, 8, 1–15.
Guttery, R. S. (2002). The effects of subdivision design on housing values: The case of alleyways. Journal of Planning Education and Research, 23, 265–273.
Hannonen, M. (2008). Predicting urban land prices: A comparison of four approaches. International Journal of Strategic Property Management, 12(4), 217–236.
Hess, D. B., & Almeida, T. M. (2007). Impact of proximity to light rail transit on station-area property values in Buffalo, New York. Urban Studies, 44, 1041–1068.
Kahn, M. E. (2007). Gentriﬁcation trends in new transit-oriented communities: Evidence from 14 cities that expanded and built rail transit systems. Real Estate Economics, 35, 155–182.
Koschinsky, J., Lozano-Gracia, N., & Piras, G. (in press) The welfare beneﬁt of a home’s location: An empirical comparison of spatial and non-spatial model estimates. Journal of Geographical Systems. doi:10.1007/s10109-011-0148-6.
Kuethe, T. H. (2012). Spatial fragmentation and the value of residential housing.
Land Economics, 88, 16–27.
Landis, J., & Zhang, M. (1995). Rail transit investments and station area land use changes: 1965–1990. In J. Landis, S. Guhathakurta, W. Huang, & M. Zhang (Eds.), Rail transit investments, real estate values, and land use change: A comparative analysis of ﬁve California rail transit systems (pp. 53–80). Berkeley, CA: University of California.
LeSage, J. P. & Pace, R. K. (2011). The biggest myth in spatial econometrics. <http:// papers.ssrn.com/sol3/papers.cfm?abstract_id=1725503> (Accessed 21.12.11).
LeSage, J. P., & Pace, R. K. (2009). Introduction to Spatial Econometrics. Boca Raton: FL,
Li, M., & Brown, H. (1980). Micro-neighborhood externalities and hedonic housing prices. Land Economics, 56(2), 125–141.
Mathur, S. (2008). Impact of transportation and other jurisdictional-level infrastructure and services on housing prices. Journal of Urban Planning and Development, 134(1), 32–41.
Matthews, J., & Turnbull, G. (2007). Neighborhood street layout and property value: The interaction of accessibility and land use mix. Journal of Real Estate Finance and Economics, 35, 111–141.
McDonald, J., & Stokes, H. (in press). Monetary policy and the housing bubble.
Journal of Real Estate Finance and Economics. doi:10.1007/s11146-011-9329-9.
McMillen, D. (1996). One hundred and ﬁfty years of land values in Chicago: A nonparametric approach. Journal of Urban Economics, 40(1), 100–124.
Oikarinen, E. (2010). An econometric examination on the share of land value of single-family housing prices in Helsinki. Research on Finnish Society, 3, 7–18.
Oliner, S. D., Nichols, J., & Mulhall, M. R. (2010). Commercial and Residential Land Prices Across the United States. FEDS Working Paper No. 16. SSRN: <http:// ssrn.com/abstract=1783627 or http://dx.doi.org/10.2139/ssrn.1783627>.
Osland, L. (2010). An application of spatial econometrics in relation to hedonic house price modeling. Journal of Real Estate Research, 32(3), 289–320.
Paez, A., Farber, S., & Wheeler, D. (2011). A simulation-based study of geographically weighted regression as a method for investigating spatially varying relationships. Environment and Planning A, 43, 2992–3010.
Paez, A., Long, F., & Farber, S. (2008). Moving window approaches for hedonic price estimation: An empirical comparison of modelling techniques. Urban Studies, 45, 1565–1582.
Pavlov, A. (2000). Space-varying regression coefﬁcients: A semi-parametric approach applied to real estate markets. Real Estate Economics, 28(2), 249–283.
Pivo, G., & Fisher, J. (2011). The walkability premium in commercial real estate investments. Real Estate Economics, 39(2), 185–219.
Rauterkus, S., & Miller, N. (2011). Residential land values and walkability. Journal of Sustainable Real Estate, 3(1), 23–43.
Ryan, B. D., & Weber, R. (2007). Valuing new development in distressed urban neighborhoods: Does design matter? Journal of the American Planning Association, 73, 100–111.
Sirmans, C. F., & Slade, B. A. (in press). National transaction-based land price indices. The Journal of Real Estate Finance and Economics. doi:10.1007/s11146-011-93063.
Song, Y., & Knaap, G. (2003). New Urbanism and housing values: A disaggregate assessment. Journal of Urban Economics, 54, 218–238.
Song, Y., & Knaap, G. (2004). Measuring the effects of mixed land uses on housing values. Regional Science and Urban Economics, 34, 663–680.
Song, Y., & Quercia, R. (2008). How are neighbourhood design features valued across different neighbourhood types? Journal of Housing and the Built Environment, 23, 297–316.
Sunding, D., & Swoboda, A. (2010). Hedonic analysis with locally weighted regression: An application to the shadow cost of housing regulation in Southern California. Regional Science and Urban Economics, 40(6), 550–573.
Tu, C. C., & Eppli, M. J. (1999). Valuing new urbanism: The case of Kentlands. Real
Estate Economics, 27, 425–451.
Wheaton, W., & Nechayev, G. (2008). The 1998–2005 housing ‘‘bubble’’ and the current ‘‘correction’’: What’s different this time? Journal of Real Estate Research, 30(1), 1–26.
Wheeler, D., & Tiefelsdorf, M. (2005). Multicollinearity and correlation among local regression coefﬁcients in geographically weighted regression. Journal of Geographical Systems, 7, 161–187.
Wiley, J., Beneﬁeld, J., & Johnson, K. (2010). Green design and the market for commercial ofﬁce space. Journal of Real Estate Finance and Economics, 41, 228–243.