Abstract
Studies have shown that proximity to light rail transit (LRT) stations positively affects property values and that these effects can appear before a system opens for operation. Here, we expand on these questions. We explore capitalization effects at several stages during the planning process for four real estate markets: single-family homes, homes in multifamily structures, commercial properties, and vacant land, using the case study of the LRT system in Phoenix, Arizona. We confirm earlier results concerning the value of proximity, and also find that markets exhibit value increases at different stages during the planning and construction process.
Introduction
In a 2001 article in this journal, Knaap, Ding, and Hopkins asked whether plans matter to urban land markets, and found that indeed planned light rail transit (LRT) construction brought positive response from markets for vacant land. We expand on this line of questioning in two ways: we refine the planning process into stages and explore how each stage affected property markets, and we do this for four kinds of markets. We remain interested in the core questions posed by Knaap, Ding, and Hopkins (2001): How does proximity to rail investments translate into values in real estate markets? Do LRT planning and construction alone, without additional changes in public policy, create these values? And if there are impacts, when during the planning process are they created?
Building on the significant work already performed in this area, we use regression analysis with interaction terms to explore changes in the value of proximity for different time periods. With long-term property sales data, we explore these issues for the recently opened LRT system in Phoenix, Arizona. Our findings confirm earlier results that most markets value positively proximity to LRT stations, and some markets value proximity to tracks negatively. We also find a significant temporal effect, showing that values from proximity to LRT, in most cases, rise with each successive planning stage.
This paper is structured as follows. We first introduce the general concepts and results from existing work concerning the relationship between property values and proximity to rail stations, focusing on LRT. Then, after describing our case study of the Phoenix LRT system, we formulate our general regression model and present specific results for each of our four property markets. The regression results are then used to show price and distance scenarios for typical properties in each property market over the successive time stages. We finish with a discussion of these results and their implications for planning and public policy.
Transportation Investments and Value
While the primary objective of public transportation investments is to improve urban mobility, they can also yield important economic benefits that are unevenly distributed. U.S. public transit agencies claim property development around rail stations as the most significant economic benefit of rail transit investments (Weinstein and Clower 2003). Investments in public transport infrastructure, such as a new rail line, are capitalized totally or partially into nearby land and housing prices (Agostini and Palmucci 2008). Theory holds that LRT might have two effects on residential property values. The first is that proximity to LRT stations might increase property values. The second is that proximity to LRT stations and tracks may decrease property values due to nuisance effects (traffic, noise, etc.). There also may be a significant secondary effect—the availability of LRT (regardless of its actual use or impact on actual accessibility) can be used to justify zoning changes or overlay districts, which allow more intense development. Many cities allow or encourage more intense development near LRT stations and this can be a source of further changes in values.
Previous Research
Numerous previous research has investigated both the spatial impacts of LRT investments as well as the timing of value impacts as systems are being planned and executed. We direct the reader to several extensive reviews of the literature, including Cervero, Ferrell, and Murphy (2002), Huang (1996), Cervero and Aschauer (1998), Landis et al. (1995), and Garret (2004). Generally, research shows that properties enjoy positive value impacts from proximity to LRT stations, though results vary based on the specific context and land-use type. Some studies also show no impacts. Most of the research uses hedonic regression or matched-pair–type approaches, which control for similar nearby properties as well as exogenous changes.
For example, a study in Portland, Oregon, showed property values increased with proximity to stations, up to 100 m from LRT stations (Chen, Rufolo, and Dueker 1997). Another study in Buffalo, New York, showed that in general, property within a ½ mile of rail stations is valued $2.31 higher (using straight-line distance) and $0.99 higher (using network distance) for every foot closer to a light rail station (Hess and Almeida 2007). A model looking at individual station effects in Buffalo revealed that the impacts are not equal throughout the system: value premiums for station proximity were greater in high-income neighborhoods than low-income neighborhoods. A study of LRT impacts on residential property values in St. Louis showed that proximity was valued at $14 per foot closer to the LRT station, for properties within about ¼ mile (Garret 2004). Atkinson-Palombo (2010) studied the effects of the planned LRT in Phoenix and found that both proximity to the planned stations and a transit-oriented zoning overlay had a significant impact on prices for both condos and single-family homes in mixed-use neighborhoods.
Some studies showed however, that in certain contexts, markets have penalized proximity to rail as a form of nuisance. For example, Landis et al. (1995) show that proximity to LRT in the case of the Santa Clara LRT had negative impacts on prices, while in the case of Sacramento and San Diego impacts were insignificant (though for San Diego, impacts were positive and significant within the central city). A later study in Santa Clara showed that home values responded positively to proximity, but condominiums did not (Cervero and Duncan 2002). Similarly, earlier studies of heavy rail in Atlanta (Nelson 1992) show that property values increased in low-income neighborhoods but decreased in high-income neighborhoods with increasing accessibility to Metropolitan Atlanta Rapid Transit Authority stations. The opposite result was found in Miami, Florida, where high-income neighborhoods benefited slightly from Miami Metro Rail but no such gains were found in low-income neighborhoods (Gatzlaff and Smith 1993).
Similarly mixed effects have been uncovered for nonresidential properties. A study in Dallas shows that value appreciation for commercial and retail properties were higher for those near LRT compared to similar properties further away (Weinstein and Clower 2003). Another study in Santa Clara County showed that properties located within ½ mile commanded premiums compared to other properties in the county (Weinberger 2001). Landis et al. (1995), however, showed ambiguous results concerning the connections between commercial property values and proximity to San Diego County LRT.
Researchers have also investigated the timing of value effects within the transportation planning process (Knaap, Ding, and Hopkins 2001; Ferguson, Goldberg, and Mark 1988; Grass 1992; Grether and Miezkowski 1974; Gatzlaff and Smith 1993; McDonald and Osuji 1995). For example, Knaap, Ding, and Hopkins (2001) found that the announcement of LRT plans and the proximity to announced station locations combined to affect significantly vacant land prices in the Portland area. Ferguson, Goldberg, and Mark (1988) found similar results in Vancouver, where value effects were measured one year before the LRT system opening and three years after station locations were announced. Another study in Chicago showed that the planning and construction of heavy rail service had significant impacts on residential land values three years before operations began (McDonald and Osuji 1995).
In this study, we consider similar questions: How does proximity to the light rail affect property values, and when do value changes, if any, occur? How do zoning changes interact with these impacts? Here, we expand the exploration of the timing of impacts: instead of pivoting around a single moment in the planning process—before or after a plan announcement or line opening—we look at value changes during the planning process divided into the environmental review, planning, construction, and operations phases. We now introduce our case study before exploring our models and results.
Context: The Phoenix Light Rail
Streetcars and rail played an important role in the development of the city of Phoenix. By the late 1800s, streetcars provided a significant part of the city’s transportation needs, lasting until 1948. As in other cities throughout the United States, conversion to buses and decentralized, automobile-oriented planning led to a relative decline in the importance of public transit. From the 1950s until the current recession, the Phoenix metro area experienced rapid exurban and suburban growth (Gober 2005). In 1999, after nearly twenty years of debate over the development of high-capacity public transportation, several cities in Maricopa County created a proposal for the Central Phoenix/East Valley Light Rail project. Supporters of LRT development argued that it would stimulate and recenter growth and revitalize downtown Phoenix and the surrounding neighborhoods. As a result of their efforts, the line now passes through central and east Phoenix and connects with the neighboring cities of Tempe and Mesa to the east. The three cities together include more than half of the county’s population of more than 4 million. Initial investment studies led to a National Environmental Protection Act (NEPA) environmental impact review process beginning in 1999, resulting in a final decision in 2003, with planning and design leading to construction beginning in 2005 (Golub, Guhathakurta, and Sollapuram 2011). Regional and local funds contributed 57 percent of the $1.4 billion cost and the LRT opened for service on December 27, 2008. There are 20.3 miles of track and twenty-eight stations.
Figure 1 shows the study area included in the 2-mile buffer around LRT stations and tracks. The system passes along what is arguably the most attractive and competitive environment for public transit in the region. Traveling west, the line begins from just west of Downtown Mesa, making stops at Arizona State University’s (ASU’s) Tempe Campus with its more than forty thousand students, in Downtown Tempe, and at the Phoenix Sky Harbor International Airport before entering downtown Phoenix. In Phoenix, the line passes by major attractors, such as the new baseball and basketball stadia, convention centers, and ASU’s downtown campus, as it turns north along the city’s historical north–south spine, Central Avenue, heading toward north Phoenix neighborhoods. Current ridership surpasses forecasts by about 50 percent: weekday ridership averaged around thirty-four thousand per day in 2009, and thirty-nine thousand in 2010, well above the twenty-six thousand projected (Valley Metro Rail 2011a). Similar experiences hold on the weekends.
Study area including 2-mile buffer around light rail transit stations and tracks
To support compact, dense, and transit-friendly development in station areas, the cities of Tempe and Phoenix developed special land-use and zoning tools such as station area plans and transportation overlay district zoning (herein TOD zoning). Phoenix was one of the first cities in the United States where station-area overlay zoning was enacted before the construction of a light rail (City of Phoenix 2002; MAG 2003). The City of Phoenix implemented TOD zoning for parcels near some of its stations in 2003, and the City of Tempe in 2005. According to the rail operator, Valley Metro Rail, close to $5.4 billion of private and $1.5 billion of public investments have been made along the corridor since 2004 (Valley Metro Rail 2011b). It should be noted that much of the investments are in downtown Phoenix and Tempe because a portion of the corridor between Tempe and Phoenix, especially as it passes near the airport, is still characterized by industrial land uses. It will take some time for residential and commercial investments to reach those areas.
The Phoenix light rail is an interesting case through which to explore issues of land use and real estate prices. It has been decades since any high-capacity passenger rail system has operated within the city, so the reaction of property markets to the announcement and construction of the system will depend specifically on perceived transportation benefits, and won’t be shaped by experiences in other parts of the same region. The Phoenix case is also interesting because there has been no previous experience in developing successful above-ground rail transit system in a hot and arid climate that limits walkability, especially during the summer months, and there was a conscious effort to transform land uses and zoning near transit stations to make them more transit oriented well in advance of the operation of the system. The Phoenix case can provide some corroboration of the benefits, if any, of coupling land use and transportation planning.
Research Approach
To understand spatial and temporal impacts on property values, we used a hedonic regression technique. Here, we discuss our regression approach and outline the general regression model that is tested and adapted for each of the real estate markets we investigated. We then present the data used in our model.
Methodology
Location theory holds that transportation investments stimulate land use activities around nodes of improved accessibility, such as around freeway interchanges or rail stations (Knight and Trygg 1977; Cervero 1984). Clustered activities result from improved accessibility to other locations both nearby and far away in the transportation network. Various statistical methods are suited to answering the question about how transportation investments may affect land values (Cervero and Aschauer 1998). To understand how property values respond to transportation planning and investments, models must be able to capture the relationship between proximity to the improvement and improvement value, as well as how that relationship might change over time. Increases in land values following the opening of a new rail facility are not necessarily caused by the new facility. Other factors like changes in the regional real estate market also have an effect on property values. The challenge is to separate transit proximity and planning process (temporal) impacts by controlling for other factors. The approach taken here is to use a hedonic regression model with interaction terms to understand how changes in real estate sale prices vary by proximity to the LRT stations and tracks, and how that relationship changes with the phases of the LRT planning process. The general model can be expressed as
where P i,t = price of parcel/property i at time t; I = structural characteristics of the parcel and property; L = spatial variables; T = temporal variables; and L · T = interaction terms between spatial and temporal variables. More detail on the specific variables used is provided below.
The dependent variable, P i,t , is the adjusted sale price of the property, calculated as the actual sale price divided by the (single-family home) Repeat Sales Index (RSI) for that month, for the city in which the property is located. 1 This operation effectively adjusts the price to constant 1989 “real estate dollars” while also controlling for exogenous market dynamics, such as the 2004-2006 housing market bubble and the subsequent recession. Using a different RSI for each city also controls for their market dynamics during the bubble and recession. Finally, because LRT planning and operations may produce different responses in the different real estate markets, analyses were preformed separately for single-family homes, units in multifamily structures (mostly condominiums), commercial properties, and vacant land.
Data
Data for this study were collected from a wide variety of sources. See Golub, Guhathakurta, and Sollapuram (2011) for more details about the data sets. The main sales price and structural characteristics were taken from the W.P. Carey School’s repeat sales database, which is used to produce the RSI. It includes all sales transactions that occurred from August 1988 to September 2010, in Maricopa County, Arizona. The private provider of the data allowed us access for research purposes. An important characteristic of this database is that it includes the actual sales transaction prices, which differ from tax assessments. The sales transactions were divided into four groups: (1) single-family homes; (2) homes in multifamily structures (townhomes and condominiums); (3) commercial properties (includes commercial or industrial uses); and (4) vacant parcels.
Temporal variables were created using a calendar of the planning process, shown in Table 1. These dates were needed to define phases for our regression, even though we recognize there is some overlap between them—for example, some planning and design activities may continue after construction has begun, etc. The last sales included in the study were from September 2010.
Phoenix Light Rail Transit Planning and Implementation Phases
We chose to confine our study area to parcels within two miles of LRT stations and tracks, given that previous studies have shown that property value impacts dissipate well before they reach the 2-mile mark. There are 122,222 parcels in the study area, with a total of more than 124,000 sales recorded in the database.
Model Development
Table 2 below presents four groups of independent variables along with the sign of their expected impact on sales price. Here, a “+” means that it is expected to have a positive effect on sales prices, a “–” means a hypothesized negative impact on prices, and “?” means we are unsure. Not all variables were used in the final regressions for each property type, because of issues related to multicollinearity. These variables are discussed further later.
Independent Variables Used in Regression Analysis (Coefficient Names Are in Parenthesis)
Sources: W.P. Carey School’s repeat sales database, publicly available Phoenix street GIS network, and 2010 Maricopa County Assessor’s “Books 100” parcel polygon GIS layer.
The structural variables help control for different property market dynamics, such as the markets for pools or two-story homes. Some are obviously linked to value, such as lot size and unit size, but others, such as bath fixtures, are proxies for number of bedrooms and may differentiate markets by household type (couples, families with children, etc.) Ideally, some categorical information about parcel zoning would be included, especially to model the vacant land markets and their reaction to the presence of LRT. Instead of using current zoning designations (of which there are more than two hundred in the City of Phoenix) as explanatory variables, we used the presence or absence of TOD overlay zoning since this was more appropriate for informing potential use and value change.
The spatial variables were constructed in a geographic information system (GIS). A GIS road data layer was used along with the 2010 Maricopa County Assessor’s “Books 100” parcel polygon GIS layer to compute distances from each parcel centroid in the study area to its nearest LRT station and track alignment. 2 Distances were produced using the “shortest network path distance” as well as straight-line distances. Freeways were excluded from this analysis as they do not provide access by pedestrians. Checks between our sales data set and County Assessor data were made to verify data quality.
Distances from significant land uses such as the central business districts (CBD) of Phoenix, Tempe, and Mesa, as well as the Phoenix Sky Harbor International Airport, were calculated for each parcel. We use these distances as proxies for likely submarkets, such as higher-priced downtown markets in the three cities, and the more industrially zoned areas close to the airport. For example, we expect proximity to the airport to be a disamenity for single-family homes. TOD overlay districts and the dates they were implemented were used to identify TOD-designated parcels and their sales.
Results
Before proceeding to a discussion and interpretation of the regression results, we must note that our statistical techniques highlight relationships between variables, and do not prove causation. Our models show that specific characteristics of properties, such as proximity to LRT stations or tracks, are strongly associated with price effects, but we don’t know for certain that decision makers (property owners or developers) were specifically basing prices on location. That said, we can interpret regression results to mean that the market, en mass, has responded to proximity to LRT and other factors by awarding higher prices for those characteristics, even though the process of causation is uncertain.
For all of the four types of properties studied here, the double-log form of the regression models provided the best fit. Checks on multicollinearity were used to ensure models were not overspecified. All linear terms exhibited low multicollinearity, while interactions terms exhibited high multicollinearity because of their multiplicative nature. The use of interaction terms always bring about questions regarding the inclusion of all constitutive (those used to form interaction terms) along with interaction terms because of multicollinearity issues. 3
Straight-line distance to tracks and stations were used instead of network distances. Pedestrian cut-throughs (especially in neighborhoods to the west of Central Avenue) were not part of our GIS road network, causing an overestimation of walking distances for some areas of the city. As a result, network distance had slightly less explanatory power in our regressions. Basic descriptive statistics for the initial data used for each model are presented in Table 3.
Descriptive Statistics
Single-Family Homes
Not surprisingly, structural variables such as square-footage, age, presence of a pool, etc. were significant in determining price (see Table 4). Temporal variables all factored positively and significantly in the price, meaning that compared to the re-NEPA review base, there was a positive value change due to the planning and implementation of the LRT project. The capitalization effect of the LRT was positive during its development and construction. TOD designation has a positive impact on property valuation. Location within 200 ft. of the tracks affects prices negatively, while proximity to LRT stations has a positive impact on prices. Proximity to the CBD affects prices positively, while proximity to the airport affects them negatively.
Single-Family and Multifamily Homes Double-Log Regression on Sale Price
While marginal significance is low, analysis of complete marginal effects show that there are significant relationships between this variable and prices at some distances from stations, or during certain time periods (see note 4). Thus, these variables will be used to generate cost curves in the next section.
Multifamily Homes
As with single-family homes, not surprisingly, structural variables such as square-footage, age, presence of a pool, etc. were the most significant in determining price for units in multifamily structures (see Table 4). Temporal variables all factored significantly in increasing prices. TOD designation showed a negative impact on property valuation, but this was statistically insignificant. In contrast to single-family homes, location within 200 ft. of the tracks affects prices positively. As with single-family homes, proximity to LRT stations and the nearest CBD has a positive impact on prices, while proximity to the airport affects them negatively.
Commercial Properties
As before, structural variables such as square footage and age, etc. were significant in determining price (see Table 5). Temporal variables all factored significantly in the price. Once again, the effect of the LRT was positive through each stage of its planning, development, and construction. TOD designation has a positive impact on property valuation. Interestingly, location within 200 ft. of the tracks affects prices positively (but minutely), while proximity to LRT stations has a positive and very strong impact on prices, compared to other factors. As before, proximity to the CBD affects prices positively, while proximity to the airport affects them negatively.
Commercial and Vacant Properties Double-Log Regression on Sale Price
While marginal significance is low, analysis of complete marginal effects show that there are significant relationships between this variable and prices at some distances from stations, or during certain time periods (see note 4). Thus, these variables will be used to generate cost curves in the next section.
Vacant Properties
Not surprisingly, lot size was most significant in determining price (see Table 5). Temporal variables show that the effect of the LRT was significant only during planning and construction. Values during NEPA review and operation did not differ significantly from the prices before the NEPA review began. TOD designation has a positive impact on property valuation. Location within 100 ft. of the tracks affects prices positively (though this effect was statistically insignificant), while proximity to LRT stations has a positive impact on prices. As in the other markets, proximity to the CBD affects prices positively, while proximity to the airport affects them negatively.
Illustrations of Price Profiles
The coefficients in log-log regressions can be interpreted as elasticities. That is, the percent change in the dependent variable induced by one percent change in the explanatory variable (dummy variables excepted). Interaction terms, however, are not as easily interpreted because of their multiplicative and additive effects. 4 Thus, to better illustrate the relationships of concern in this piece that include the effects of distance, time, and TOD designation on prices, we constructed several distance scenarios for the “mean property” for each of the four property types. Here, the mean values 5 are taken for all inputs other than distance to LRT stations. The distance to LRT stations for that property is then varied and regression coefficients, including the interaction terms, are used to predict the price per square foot of the property. Each scenario is presented separately below for each property type in 2010 “real estate dollars.”
Impacts of Distance to LRT on Single-Family Homes
Figure 2 illustrates the impact of distance to LRT stations on single-family home prices for a house with the mean features, during the different time periods. Note the decrease in value within 200 ft. results from the penalty of being close to the tracks—a “nuisance” effect. Alternatively, the spike in value after 200 ft. results from the accessibility benefit of being close to the stations. For example, before the NEPA review began, the mean unit of 1,425 sq. ft. on a 7,590-sq.ft. lot (slightly less than 1/6 acre), would sell for around $111,000 (about $78/sq. ft.) located 200 ft. from the LRT station and around $105,000 (about $74/sq. ft.) at 10,000 ft. from the LRT station. This relationship changes during each time period. During the NEPA planning and design and construction periods, the premium for proximity to the LRT rises with each period. The operation phase experiences drop in prices for all properties in the study period compared to earlier periods, countered by an even more pronounced premium for proximity, effectively preserving prices for properties near LRT stations. This makes the slope of prices with distance much more pronounced than in earlier periods. During the operation phase, the value of the average unit rises to about $101/sq. ft. at 200 ft., and falls to about $54/sq. ft. at 10,000 ft.
Impacts of distance to light rail transit on single-family home prices
Impacts of Distance to LRT on Multifamily Homes
Here, proximity to an LRT station is a benefit to prices and there is no nuisance effect (within 200 ft. of tracks; see Figure 3). For example, before construction began, the mean unit of 1,118 sq. ft. would sell for around $81,000 (about $73/sq. ft.) located 200 ft. from the LRT station and around $75,000 (about $67/sq. ft.) at 10,000 ft. from the LRT station. This relationship changes in subsequent time periods as prices rise pretty uniformly for all parcels in the area, while during the operation phase, prices drop for properties further from the LRT, while those closer gain value above earlier periods. Similar to single-family homes, the resulting slope of prices over distance is more pronounced than in earlier periods. During the operation phase, the value of the average unit rises to about $100/sq. ft. at 200 ft., falling to about $65/sq. ft. at 10,000 ft.
Impacts of distance to light rail transit on multifamily home prices
Impacts of Distance to LRT on Commercial Properties
Note the slight decrease in value within 200 ft. results from the nuisance effect, but was only significant during the construction period (see Figure 4). The regression model predicted fairly similar prices during the Pre-NEPA, NEPA, and planning and design phases. During planning and design, the mean unit of 14,958 sq. ft. would sell for around $730,000 (about $49/sq. ft.) located 200 ft. from a LRT station and around $454,000 (about $30/sq. ft.) at 10,000 ft. During construction, prices increased fairly uniformly for all properties in the study area, though exhibiting a similar slope of price with respect to distance to LRT stations and nuisance effect from proximity to tracts (and presumably construction nuisances). Prices rose again during the operations phase, with a slightly steeper slope (with no nuisance effect), and the value of the average unit rises to about $85/sq. ft. at 200 ft. and falls to about $37/sq. ft. at 10,000 ft.
Impacts of distance to light rail transit on commercial property prices
Impacts of Distance to LRT on Vacant Properties
Proximity to the LRT station is a benefit to prices and there is no nuisance effect (within 200 ft. of tracks; see Figure 5). For example, for time periods other than during construction and planning and design, the mean vacant parcel of 24,198 sq. ft. would sell for around $150,500 (about $6.22/sq. ft.) located 200 ft. from an LRT station and around $82,000 (about $3.39/sq. ft.) at 10,000 ft. from a LRT station. This relationship changes during the construction and planning and design phases. During the construction period, the value of the average vacant parcel rises to $16.62/sq. ft. at 200 ft. and falls to about $2.39/sq. ft. at 10,000 ft. According to the model, prices during operations return to levels similar to those found before the planning and design period.
Impacts of distance to light rail transit on vacant property prices
Impacts of TOD Designation
Taking the mean units and looking at the impact of the TOD coefficient on predicted prices provides an illustration of the impact of TOD designation on values. TOD designation only affected prices for single-family homes 6 and vacant parcels. For the average properties sold for each type, TOD designation added 56 percent to values of vacant parcels, and 6.8 percent for single-family homes.
Discussion and Conclusions
We can now reflect on the original questions that motivated this work: how does proximity to rail investments translate into values in real estate markets? Does LRT planning and construction alone, without additional changes in public policy, create these values? And if there are impacts, when during the planning process do they emerge? This study corroborates previous work regarding the first two questions: it shows a positive value impact from proximity to LRT in general and sometimes the significance of nuisance effects from proximity to the tracks. It shows that there is an effect of proximity to LRT, controlling for other public policy changes, such as transit oriented rezoning, in several different property markets. Finally, results show that positive value effects happen well before the operations of the system begins and can accrue steadily throughout the planning and construction processes.
Prices for single-family homes responded significantly to both proximity and planning phases. Proximity to LRT translated into prices which increased in each successive time period. During the operation period, properties away from LRT lost value compared to earlier times. It should be noted that operation began only months after the housing market recession, which hit Arizona and the Phoenix area particularly hard. Though we did control for citywide home prices, there may be other issues for which we could not control, leading to the fall in prices during operations. Nonetheless, for an average property at 200 ft. from LRT, values increased during each phase, culminating in a price after operation of around 25 percent higher than prices before NEPA review began. Furthermore, prices did not vary much by proximity before LRT was announced, as would be expected. In addition, there was a nuisance effect even before LRT announcement, likely because the LRT alignment is mostly along arterial roads where there may already exist a penalty for single-family home values. Interestingly, this nuisance appears to lessen in severity over time. Regarding TOD designation, we found a mild impact on values.
Homes in multifamily structures showed a similar rise in prices for properties very close to LRT, a relationship that grew more pronounced over time. Still, prices for properties far from the LRT dropped during the operations phase—likely showing the effects of the general real estate market recession not captured in the housing price index used to control for general price changes. Proximity to the LRT tracks, normally expected to produce nuisance effects and lowered prices, did the opposite; prices were higher for properties within 200 ft. of LRT tracks. This may be due to the prevalence of several new multifamily structures along the LRT alignment.
Commercial properties show a similarly positive effect of proximity to LRT, increasing in value with each successive time period. Unlike the two residential markets, there is no general fall in values after operations begin, revealing that perhaps the commercial real estate market decline is better captured in the real estate price index used to adjust prices for marketwide conditions.
Finally, vacant land shows a positive impact from proximity to LRT as was expected based on a review of other studies in this area. The value effects over time are somewhat puzzling: there is only a significant (positive) effect during the planning and design and construction periods. During the construction period, there is a pronounced impact of proximity, with a slight decrease for properties further away. This may reflect the heating up of the vacant parcel market just before the LRT operation phase (2005-2007), which coincides with the general real estate bubble. TOD zoning had a significant effect on values for vacant parcels.
At this point, we can reflect on these results in the light of two potentially significant background processes. The first, the housing bubble, caused prices to peak during the 2005-2007 period (the LRT construction period) and clearly affected results for single-family and multifamily homes. It is important to note that soon after the collapse of the housing market and after the system started operating, the premium for LRT access actually increased in most cases for properties very near station areas. However, these premiums declined rapidly away from a station and fell below the prices that the properties were commanding prior to operation. The stark contrast between the general fall in housing prices citywide during operations phase due to the housing market crisis, and the maintenance of prices for properties very close to stations is significant. This finding offers strong evidence of the positive capitalization effects of LRT access.
The second background process of significance here is the general development bias inherent in the placement and planning of the state’s first light rail corridor. This corridor contains many prime development sites for the region, and would arguably be the focus of investment priorities, and thus price improvements, with or without light rail. The question then becomes, can the regression separate out effects of LRT from this background value shift? It is impossible to know at what scale this background value accrues, and so we must keep in mind that some of the price effects from proximity to LRT may indeed reflect those general corridor development and investment processes. However, it is notable that premiums for LRT access in most cases increased, with each phase of development strongly suggesting that the system did have a capitalization impact on adjacent properties.
Despite potentially confounding effects of the real estate bubble and any background corridor development process, there are some clear and important results from this work. We find that value changes can occur throughout the planning period, and that overwhelmingly, proximity yielded positive value benefits. Occasionally, nuisance effects were significant but not across all cases. Single-family homes showed nuisance effects, while multifamily homes showed value increases near LRT tracks. Commercial properties showed a nuisance effect only during construction.
Based on this study, we can answer the question posed by Knaap, Ding, and Hopkins—“Do Plans Matter?” We find that they matter. We also show that completing environmental review adds additional value, as does breaking ground for construction, as does the opening of the system. Our research shows that there can be increments of value accretion as a project is seen through its various planning stages, and that different property markets respond differently.
These results have important public policy ramifications. Understanding first that property markets value proximity to LRT, and second, when these valuations seem to occur can help cities plan property acquisitions or rezoning during the process of LRT planning. This research points strongly to the need for cities to do their property purchases or assemblies as early in the planning process as possible in order to take full advantage of the valuation increases they are causing through their actions. The earlier they get into the market, the more valuation increases they can capture for their public investments for land, transportation, affordable housing, or other developments. In a broader sense, tax assessments or other value capture policies should take into account the timing of valuation changes vis-à-vis LRT planning. Finally, those concerned about displacement effects of value changes due to LRT need to get involved early in the process to secure affordable housing opportunities before land or properties have grown in value.
Footnotes
Acknowledgements
This work relied on administrative support from the Design School and the School of Geographical Sciences and Urban Planning at Arizona State University. It also benefited greatly from the insights of our project managers at the Maricopa Association of Governments and from data support from Ion Data and the Center for Real Estate Theory and Practice at the W.P. Carey School of Business at Arizona State University. The authors wish to thank the editor and anonymous reviewers for their careful review and valuable comments.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was funded by a grant from the Maricopa Association of Governments [Contract 382].