Hedonic Evaluation of the Effects of EmX Routes on the Value of Commercial and Mixed Use Properties

Transcription

1 Hedonic Evaluation of the Effects of EmX Routes on the Value of Commercial and Mixed Use Properties By Eli Goodwin and Zack Snyder Presented to the Department of Economics, University of Oregon, as partial fulfillment of requirements for honors in Economics, under the supervision of Professor Joe Stone. June 13, 2013 Abstract: Investment in public and mass transit can be viewed as amenities for adjacent and proximal businesses and commercial properties thereby increasing their sale value. Bus rapid transit is a cheaper alternative to light and heavy rail, requiring fewer modifications of existing transit infrastructure. Lane Transit District (LTD) located in Eugene/Springfield, Oregon has operated a BRT system known as the Emerald Express (EmX) for the past 6 years. Currently, LTD has plans to expand service to West 11 th, but has met harsh criticism from local businesses located along the proposed extension, stemming from concerns regarding revenue disruption, decreases in property value, and increases in congestion. The effect of the presence of the EmX and associated stations on local commercial and mixed use property sales along the Franklin corridor was tested and was found to have a positive base effect on property values in addition to an effect that was distance dependent within the corridor. These effects were used to forecast gains in commercial property value that may be realized along the West 11 th corridor. It was then found that commercial and mixed-use properties along the West 11 th corridor are expected to see gains in property value due to the proposed extension but at a depressed rate relative to the rest of Eugene.

2 Table of Contents I. Introduction A. Description of BRT B. LTD BRT Implementation: EmX C. Literature Review D. Hypotheses II. Methodology A. Question B. Data Specification C. Variable Specification D. Model Specification III. IV. Results Discussion V. Conclusion VI. VII. Works Cited Figures VIII. Tables

3 I. Introduction A. Description of BRT In cities with established transportation infrastructure congestion relief can be problematic due to unwelcomed compromises between traditional transit systems and congestion solutions commuters do not want to sacrifice automobile lanes for bike lanes or heavy rail or other inconvenient systems. Bus rapid transit (BRT) is presented as solution to congestion relief in metro areas, is cheaper to introduce than traditional rail systems and is compatible with currently existing infrastructure with some modification and without complete sacrifice of automobile lanes. BRT describes various types of bus transit systems that combine elements of light or fixed rail transit (LRT) with elements of familiar bus transit systems, synthesizing the flexibility of bus transit systems with the efficiency of LRT. BRT buses can operate on normal roadways and/or in dedicated bus lanes, allowing greater flexibility in route design and lower implementation costs than LRT. Because BRT often operates in dedicated lanes and is often given traffic signal priority, it is able to bypass congestion and decrease transit time normally faced by traditional bus transit, hence why it is considered rapid transit. BRT busses are also designed differently than traditional buses BRT buses have station level floors, sliding doors, and articulation allowing for greater passenger volume and faster passenger loading and unloading. All of these elements combined create an efficient and cost effective transit solution for cities with congestion problems [1].

4 B. LTD BRT Implementation: EmX Transit planners differentiate BRT lines form already implemented bus transit by branding their BRT systems; in Lane County, LTD brands their BRT line as the Emerald Expressway or EmX. Due to concerns regarding congestion, LTD began planning a BRT line in 1996 in order to decrease transit time along key corridors in the Eugene Springfield area and increase ridership and in 2001 the Franklin route was approved. The first EmX route was the Franklin corridor and was selected due to the dense population, proximity to the University of Oregon and Sacred Heart Medical center, and would serve as the backbone of LTD by rapidly connecting the two largest stations in the transit district. LTD completed construction of the first EmX line in early 2007 connecting the downtown Eugene LTD station to Springfield station along W 11 th, Franklin, and Main (Figure 1) [2]. The route saw ridership more than double in the first year and in the years following saw on average 1.6 million boarding per year, proving to be very popular with commuters [3]. The first EmX extension project was a route connecting the Springfield station to Riverbend Hospital, Pioneer Plaza, Gateway Industrial Park, and Gateway Mall; construction was completed in Ridership along the Gateway route has increased since implementation with 900 thousand annual riders [4]. The next proposed extension was a route connecting the downtown Eugene station to the intersection of W 11 th and Randy Pape Beltline Highway. The West 11 th EmX extension is proposed as a solution to the congestion problems along the corridor; in years past, there was a push for the building of a west Eugene expressway, but the West Eugene community has opposed its construction out of

5 concern for environmental degradation, wetlands destruction and disruption of neighborhoods (Figure 2) [5]. Currently, there is no way to bypass the congestion along West 11th if one is travelling from East/South Eugene to West Eugene, other than traveling on Randy Pape Beltline Highway. With the push for a West 11 th extension, there has been significant opposition to the construction of the extension by businesses located along the proposed route. Opposition is being represented by a group called Our Money, Our Transit (OMOT). OMOT objects to the project due to concerns of obstructed access of commercial properties caused by route construction, cost of construction and upkeep of the EmX services, lack of demand for BRT along the corridor, businesses permanently shutting down, increases in congestion due to the EmX, and land purchases for the construction of EmX lanes. The vocality of OMOT in their opposition suggested businesses were facing significant costs due to construction. The initial aim of this study was to estimate the short-term effects on revenue a business would face during construction and how the EmX would impact future revenue by measuring the impact EmX had on businesses located along current EmX routes. Unfortunately, revenue data for businesses along the current and proposed routes is unavailable. The focus of this study has since shifted to measuring the impact EmX has on commercial properties; specifically, determining what effect does proximity to an EmX station have on the value of commercial properties along the Franklin EmX route. The effect of EmX on commercial property values is critical for assessing whether or not the EmX projects are beneficial to landowners who are affected by route construction and if EmX influences patterns of development in the Eugene and Springfield Metro area. If EmX does negatively impact commercial property

6 owners, LTD must implement strategies to compensate property owners for losses or find ways to mitigate future negative impacts. C. Literature Review There is growing research detailing the effects of BRT on property values in North America as implementation of BRT in metro areas increases [6]. Much work has been done on the effect of light and heavy railway stations on property values, with mixed results. The majority of these studies use some form of a hedonic price model to explain the variation in property value in relation to proximity to railway stations, using distance, property attributes, lease, real estate transactions, and/or assessed tax value data [6,7]. Observed effects vary by study and location; some studies see a direct negative correlation between proximity and residential property values while others report a weakly positive correlation between proximity and property values, indicative of a nuisance effect. One study observed a sweet spot effect on residential property values, i.e. not too close for the station to be a nuisance and but close enough that the station could be considered an amenity [7,8]. A case study that came out of Santa Clara in 2001 is of particular interest due to parallels to the current situation on W11th in Eugene. Commercial property owners were opposed to the construction of a LRT line and associated stations because they believed it would harm their property values. Commercial property owners eventually filed suit against the county for damages after the line was completed and in service. Using a hedonic price model and lease information while controlling for other factors such as highway access, property improvements, and economic conditions, commercial

7 properties located along LRT routes and proximal to stations demanded higher rents through than those outside of the LRT corridor [9]. Last year, two University of Oregon students, Hodel and Ickler, analyzed the effect of EmX stations on residential property values along the Franklin and Gateway EmX route using the RLID data set. The RLID data set contains real estate transaction data from Lane County and includes geocode data, amenity information, date of sale, and transaction value and will be discussed further in the following sections of this paper. By controlling for distance to various key places in the Eugene Springfield area and amenities of the property, Hodel and Ickler observed that proximity to an EmX station caused a.18 % price increase if a property was within one minute of walking distance to a station [10]. Using their study as a template, this paper serves to estimate the effects of EmX stations on commercial and mixed-use property values along the Franklin route. D. Hypotheses Though there have been few studies conducted on the economic impact of bus rapid transit, there is an ample amount of evidence available detailing the economic effects of LRT on development. Based on previous research completed in this area in addition to Hodel and Ickler s 2012 paper, it is hypothesized that the EmX has a positive effect on commercial and mixed use property sale values in the Franklin corridor and that this effect is nonlinear and inversely distance dependent. That is, as distance to an EmX station increases value of commercial and mixed-use properties decreases with properties located directly on the route experiencing a greater increase than those located a quarter or half-mile off the route. We specified the Franklin corridor as all properties located within a half-mile of a current EmX station along the Franklin route. Properties outside

8 of this radius were expected to experience no effects from the route and thus serve as the control group. Finally, if both the Franklin Boulevard corridor and West 11 th corridor have experienced similar trends in property values prior to Franklin EmX completion, then trend effects due to the implementation of the EmX on the Franklin corridor can be used to forecast possible effects on West 11 th commercial and mixed-use property values due to the completion of the proposed EmX extension on West 11 th. II. Methodology A. Question The central purpose of this study is to evaluate the effect the presence of the Franklin EmX route has had on the value of commercial and mixed use properties along the Franklin Boulevard corridor as reflected by changes in market value over the past six years. Directly following, this study serves to forecast possible changes in the value of commercial and mixed use properties along West 11 th Avenue that may be realized by the implementation of the proposed West 11 th EmX extension. Hedonic modeling techniques were utilized to evaluate commercial and mixeduse property values. Developed in Lancaster s 1966 Theory of Consumer Demand and later in Sherwin Rosen s 1974 paper, hedonic prices are defined as the implicit prices of attributes and are revealed to economic agents from observed prices of differentiated products and the specific amounts of characteristics associated with them [11]. 1 That is, property value can be seen as a function of the aggregated individual effects physical characteristics and amenities have on a given property. At this point it is important to recognize that because this study only establishes the effect of one of the physical 1 Hedonic valuation was developed primarily by Sherwin Rosen in his 1974 paper titled Hedonic Price and Implicit Markets: Product Differentiation in Pure Competition. 2 Roughly twenty percent of the data available at the Lane County Tax Assessment Offices was

9 characteristics of a given property total acreage hedonic valuation techniques are not being utilized in the purest sense. B. Data Specification Two large sets of data were considered when composing this paper: the Regional Land Information Database (RLID) of Lane County and Property Tax Assessment data from the Tax Assessment Offices of Lane County. RLID details all market transactions of commercial, mixed-use and residential properties over the past twenty-five years. Observations in this set are organized by sale year and price in addition to the latitude and longitude coordinates of the property. The zoning restrictions and physical characteristics of some of the observations are also included in the data. The latter set of data from the Tax Assessment Offices of Lane County was not immediately accessible and after review proved unsatisfactory for the purpose at hand, leading to the employment of only the RLID set in the final regressions and analyses. 2 The initial plan was to determine and differentiate between the individual effects the presence of the Franklin and the Gateway EmX routes have had on properties in their respective areas. However, it quickly became evident that observations from the Gateway area would not prove useful due to the fact that the Gateway EmX route was only recently completed in 2011 and only twelve commercial and mixed-use properties in the area sold since the completion of the route. For this reason, observations from the Springfield and Gateway areas were dropped and efforts were focused on Eugene commercial and mixed-use properties, in particular those along or near the Franklin Boulevard corridor. After dropping the Springfield observations and accounting for 2 Roughly twenty percent of the data available at the Lane County Tax Assessment Offices was digitized, requiring the remaining eighty percent of observations to be analyzed by means of paper and pencil.

10 outliers, the data in the RLID set contains 1,824 observations and takes on the distribution illustrated in Figure 3. Notice that the distribution is positively skewed, an observation that sometimes is consistent with heteroskedasticity, though not the case in this study. 3 Though the RLID set proved much more useful than the Tax Assessment data, it still has its shortcomings. Foremost, the RLID data is incredibly inconsistent for commercial and mixed-use properties. The physical characteristic variables simply are not available for a very large majority of the observations. 4 For this reason the only variable present that accounts for the physical layout of a given property considered when running regressions was total acreage. 5 Though acreage plays a substantial role in determining property values, it is important to recognize that the shortcomings of the data and therefore the shortcomings of the models may influence the significance of the results. In addition to inconsistencies in the data, commercial properties do not change ownership via market transactions every year. This fact presents a serious problem in that a given property that sold in 1986 may be physically and fundamentally different than when it sold in Moreover, the data did not control for overall changes in the economy nor did it control for inflation, phenomena required to account for. 6 In addition, Google maps was utilized to determine latitude and longitude coordinates of certain locations in and around Eugene other than the commercial and 3 A Goldfeld- Quandt test was performed and found no significant signs of heteroskedasticity. 4 Moreover, all zoning variables proved insignificant in explaining sale value. 5 The initial plan was to account for the other physical attributes of a property by adding a variable controlling for fixed property effects between sale years. However, the majority of the properties sold only once in the past decade, making it nearly impossible to control for fixed property effects for all observations. 6 Sale prices were adjusted using the consumer price index in order to account for inflation over the past twelve years.

11 mixed-use properties detailed in the RLID set. These locations include major intersections and points of interest that are detailed in Figure 4. C. Variable Specification Hedonic valuation techniques aggregate the individual effects of exogenous variables on a property s value in order to determine the overall value of a given property. Accordingly, it is assumed Eugene commercial and mixed-use property values are a function of four groups of variables as follows: physical characteristics of the property, proximity to major points of interest in the area, presence of and proximity to an EmX route or station and general economic trends in the market. As discussed previously, the available data significantly hinders the ability to utilize physical characteristics as determinants of property value. The only variable of this type that remains consistent across all observations is total acreage and is assumed to have a substantial positive effect on the value of commercial properties. The remaining exogenous variables present in the following models were developed in order to account for determinants of property value other than the physical state of the property. Ideally, this subgroup of variables increases the explanatory power of the models and combats inconsistencies present in the data. A set of variables was then developed to capture distances between observations and current EmX stations. A Stata macro titled Traveltime developed by Ozemik and Miles in 2012 was initially to be used, however after contacting the authors it was discovered that the outdated Traveltime script did not have the ability to employ the latest edition of Google maps. As a remedy, a MATLAB script that served the same purpose as

12 Traveltime was developed. 7 Data produced in MATLAB was then exported to Stata for manipulation and regression. The result was a group of variables that took on the value of the distance (km) between observations in the data set and certain points of interest in Eugene: Autzen Stadium, the Erb Memorial Union at the University of Oregon, Lane Community College, Gateway Mall, Valley River Center, Oakway Center, Costco, Sacred Heart Hospital, a list of several prominent intersections in and around the Eugene area and a list of current and proposed EmX stations along the Franklin and West 11 th routes, respectively. Directly following, a variable was generated taking on the value of the distance to only the nearest EmX station from each observation, titled dist_nearest_station. 8 The final set of variables was produced to control for the completion of the Franklin Boulevard EmX route and for general market trends. 9 A dummy variable titled franklin_route that takes on a value of one for all observations sold after the year 2005 and a value of zero for all observations sold during or prior to the year 2005 was utilized in addition to a variable titled trend that takes on the value of the year in which a given property sold less two thousand in order to account for the business cycle The script employed the Haversine formula in order to account for the shape of the surface of the Earth when calculating the distance between two points. 8 For instance, a given property that is located 0.1 km from Hilyard Station but more than 0.1 km from all other stations along the route will produce dist_nearest_station = Though the Franklin route was completed in January 2007, it is reasonable to assume the presence of the future EmX route was taken into account when determining sale prices for properties that sold in the year Moreover, initial regressions that assumed 2007 as the first treatment year generated sporadic and unrealistic results, possibly due to interaction between the presence of the EmX and the 2007 Financial Crisis. 10 For instance, a property that sold in the year 2005 is associated with a trend value of five.

13 D. Model Specification The models developed in this paper are cross sectional time series analyses that model changes in commercial and mixed-use property value due to the presence of an EmX route as reflected by market transactions that took place between the years 2000 and The inconsistent nature of the data in addition to the small number of observations available required specifying multiple models in order to get a more accurate grasp of the magnitude and significance of the effect the EmX route has had on commercial and mixed-use property values over the past six years; by specifying and regressing multiple models it became clear which of the variables possessed significant explanatory power in order to build a more robust, final hedonic model. The first task was to determine any overall changes in property value experienced by properties within the test group due to the implementation of the Franklin EmX route and whether or not this effect was being realized over time. Accordingly, the preliminary model is as follows: ln(sale_ price) = β 1 + β 2 ln(acreage)+ β 3 half _ mile + β 4 franklin _ test +β 5 trend + β 6 trend 2 + β 7 trend _ corr _ premx + β 8 trend _ corr _ emx + u i Note this model employs a log-log specification in that the natural log of both the endogenous and an exogenous variable was employed in order to illustrate changes in property value due to the EmX in percentage terms. The variable half_mile is a dummy variable that takes a value of one for all properties located within the test group. The variable franklin_test is an interaction term that is produced by interacting the half_mile

14 and franklin_route variables. 11 The first crucial takeaway of this model is the coefficient associated with this variable; this coefficient reflects the overall change in property value experienced by properties along the corridor due to the presence of the EmX. The other main variable at play here is trend. This variable was interacted with itself in order to determine any exponential changes in commercial and mixed-use property values due to the progression of time, producing the variable trend_squared. The second important takeaway of this model is the difference in the coefficients associated with the variables trend_corr_premx and trend_corr_emx. The prior is produced by interacting trend and the half_mile dummy variable for all observations transacted in the market prior to The coefficient on this variable measures the effect of market trends on the value of commercial and mixed-use properties along the corridor prior to the implementation of the EmX. The variable trend_corr_emx captures the same effect but only on those properties that were sold after the year The difference in the coefficients associated with these two variables measures the change in the effect of market trends on commercial and mixed-use property values along the corridor due to the implementation of the EmX. The second model in this study employs hedonic valuation techniques more so than the previous model and can be used to illustrate changes in commercial and mixeduse property value as a result of being located within a certain proximity to an EmX station. This hedonic model incorporates aspects of the previous model in addition to two new interaction terms. This model also employs distances between 11 This variable takes a value of one only for those properties that sold after the year 2005 and are located within a half- mile of a current EmX station.

15 commercial/mixed-use properties and certain points of interest in Eugene as additional exogenous variables. The hedonic model is as follows: ln(sale_ price) = β 1 + β 2 total _ acreage + β 3 half _ mile + β 4 franklin _ test +β 5 trend + β 6 trend 2 + β 7 corr _ dist _ premx + β 8 corr _ dist _ emx + β i d i The only differences between this model and the previous model are the exclusion of the trend interaction terms, the specification of the model (now of a log-linear form), and the inclusion of the last three terms: dist_corr_premx, dist_corr_emx and the group of variables labeled d i. The variable titled dist_corr_premx is an interaction term that takes on the distance from a given property to the nearest EmX station along the Franklin route for all properties that sold prior to the year 2006 that are located within a half-mile of an EmX station. The second term, dist_corr_emx was generated using the same process as the previous variable, but only for those properties located within a half-mile of an EmX station and that sold after the year The difference between the coefficients associated with these two variables represents the change in property value due to the implementation of the EmX as a function of proximity to a station along the route. Finally, the term labeled d i is a set of generated distance variables that take on the value of the distance between a given commercial property and the major points of interest in Eugene detailed in the previous section. This final set of variables was included in order to give the model more explanatory power and to combat the inconsistencies of the RLID data set, specifically that being located near a highly

16 trafficked area can be viewed as an amenity of a commercial property. It follows that this model is indeed hedonic in nature and should be interpreted as such. III. Results The results of the base and trend effect model which tested the effect of logacreage, presence in the Franklin corridor, general sale trends in Eugene, gain in value post EmX completion along the corridor, and trend along the corridor pre and post EmX completion on log-sale price are summarized in Table 1, column 1. All regressors were significant to at least the p<0.05 level. Being located on the corridor pre and post EmX completion had positive effects on sale value and the general sale trend in Eugene was positive, however the coefficients associated with trend_squared, trend_corr_premx, and trend_corr_emx were negative. That being said, trend_corr_emx was less negative than trend_corr_premx indicating that property values along the corridor were still decreasing after the implementation of the EmX, but at a lesser negative rate than what would have been experienced had the EmX route not been implemented. The results of the hedonic model which tested for the effects of distance to various amenities and intersections, presence in the Franklin corridor, presence in the corridor post EmX completion, acreage, sale trends in Eugene, distance to nearest EmX station pre and post completion along the corridor on log-sale price are summarized on Table 1, column 2. Of the distance and intersection amenity variables, Autzen Stadium, U of O, LCC, Delta Oaks, Beltline and 11 th, Beltline and Barger, Beltline and Royal, Chambers and 18 th, Oakway Center, Sacred Heart Hospital, and Willamette and 40 th were significant, with a mix of positive and negative coefficients. Other intersections and amenity variables had less to no significance with different signs attached to their

17 respective coefficients, but distance and amenity variables with no significance were retained in the model because of their effect on R 2. Trend in the hedonic model did not change in significance from the previous trend regression, remaining significant to at least the p<0.01 level. Presence along the Franklin corridor did not change in significance relative to the trend and base effect model. The coefficient associated with the franklin_test variable, denoting a property that sold in the corridor post EmX completion, reduced in significance but not magnitude relative to the base and trend effect regression and remained positive. Pre and post EmX completion distance to nearest EmX station interaction variables were both highly significant and negative, with the post EmX completion distance interaction variable indicating an increase in the cost of being located off the Franklin route. The general trend regressions of West 11 th and Franklin corridors are summarized in Table 2. In both the West 11 th and Franklin regressions, the coefficients associated with ln_acres, trend, and trend_squared were highly significant and were similar in magnitude. For the Franklin corridor, trend pre and post EmX completion were positive with post EmX completion only being slightly greater. Trend along the West 11 th corridor was negative relative to the rest of Eugene. IV. Discussion The first regression, referred to as the base effect and trend model, reported coefficients that were significant at the p<0.05 level and signs attached to coefficients for ln_acres, half_mile, franklin_test, and trend were positive, as expected. The coefficient reported on franklin_test (0.415) represents an initial gain in property value of 41.5% after EmX completion, a result that is perhaps unrealistic at first glance. Notice the

18 coefficients attached to trend_corr_premx and trend_corr_emx were negative and significant, indicating that property values were declining over time along the corridor prior to and after the EmX completion. Looking at the difference in the coefficients associated with trend_corr_premx (-0.117) and trend_corr_emx (-0.059), property values increase by 5.8 percentage points in trend effects within the corridor post EmX completion. That being said, property values are still trending at a depressed rate per year relative to the rest of Eugene. Combining the effects of franklin_test and trend_corr_emx, properties sold in 2006 see an initial increase of 6.1% in value, however value declines each year thereafter due to the overall negative trend effect on the corridor. 12 That being said, this annual decline in value is still less than the decline in value properties would have experienced if the EmX had not been constructed. The hedonic model reports similar effects as the base effect and trend model, though the magnitudes of the effects differ. The coefficients reported on total_acres, half_mile, trend, trend_squared, corr_dist_premx and corr_dist_emx are all significant at the p<0.1 level. Again, the coefficient reported on franklin_test (0.414) was not only significant but also illustrates a large gain in property value due to the implementation of the EmX consistent with the unrealistic effect reported in the base effect and trend model. In addition to base effects, the hedonic model measures the effect of proximity to an EmX station on the value of commercial and mixed-use properties. Looking at the difference in the coefficients associated with corr_dist_premx (-0.826) and corr_dist_emx (-1.633), property values are declining at a rate that is 80.7 percentage points per kilometer off the route more than if the EmX had not been constructed. This represents a 12 Calculated by multiplying the difference in the corr_dist_premx and corr_dist_emx coefficients by 0.8 (one half- mile) and subtracting this number from the coefficient associated with franklin_test.

19 substantial and implausible decrease in the property value for properties that are located within the Franklin corridor. By combining the effects of the coefficients associated with franklin_test, trend_corr_emx and corr_dist_emx a more realistic measure of the overall effect of the EmX on property values can be discerned. This model predicts a property located a half-mile off the route experienced a decrease in property value of 23.1% once the EmX was completed. However, the model also reports an increase in the value of a property located a quarter-mile off the corridor of 9.2% due to the EmX.[footnote] Not only is this consistent with the hypothesis that the effect of the EmX on property values within the corridor is nonlinear, this result also illustrates the effect of the EmX is dwindling in properties near the edge of the half-mile corridor. Again it must be stressed that the magnitude of this effect is likely be inflated due to limitations of the data, but at the very least there seems to exist a positive effect on the value of commercial and mixed-use properties within the corridor due to the implementation of the EmX. It follows that properties located along the West 11 th corridor may experience similar effects due to the proposed EmX extension, though the magnitude of these effects needs to be interpreted carefully. The trend regressions for the Franklin and West 11 th corridors detailed in Table 2, columns 1 and 2 respectively, reported high significance for regressors. The value of the coefficients shared between the two regressions differ by 1-2 percentage points, as expected because the only difference in regressors between the two regressions is the corridor of interest and the post EmX effects along the Franklin corridor. The effects of EmX completion have not yet been observed on the West 11 th corridor, but have been

20 observed on the Franklin corridor and EmX effects along the Franklin corridor can be extrapolated to the West 11 th corridor. Looking at Table 2, the difference in trend_corr_premx and trend_corr_emx along the Franklin corridor represents a gain of 1.23 percentage points. Applying this change to the current West 11th corridor, the effect of trend on property values along the West 11 th corridor post EmX completion is expected to be -2.13% per year. That is, commercial and mixed-use properties along the West 11 th corridor will potentially experience a change in property value of -2.13% per year after the implementation of the proposed EmX extension, whereas properties along this corridor have been and are currently experiencing a larger change in property value of -3.36% per year. However, looking at summary statistics detailed in Table 3, there are differences in the volume of commercial and mixed use properties sold along each corridor with West 11th having only roughly 1/5 of the properties sold along the Franklin corridor since the year There are also differences in the mean sale prices and mean acreage of properties along both corridors, specifically the mean sale price of properties on West 11 th corridor sold in the past 12 years is ~90% of the mean sale price of those that sold on the Franklin corridor. Additionally, the mean acreage of properties sold on the Franklin corridor is ~1/3 of the mean acreage of properties sold on the West 11 th corridor. These differences raise concerns about the accuracy of forecasted effects and of course the possible variable biases that are present throughout the regressions performed in this analysis. V. Conclusion In order to determine the effect of the EmX and associated stations on commercial and mixed used property values along the Franklin Corridor we used basic trend and

21 hedonic regressions and observed a positive and non-linear effect on commercial and mixed use property values. Specifically, prior to EmX completion properties along the corridor were depressed in value relative to property sale trends in the Eugene area, but after completion of the EmX property values were still depressed, but were trending at a less depressed rate than what would have been experienced had the EmX not been constructed. The hedonic model demonstrated that property values in Eugene could be explained by distance to various amenities and intersections, total acreage, and general property sale trends in Eugene. Along the Franklin corridor pre EmX construction, it was observed that presence within the corridor had a positive effect on property value relative to other locations and that there was a distance penalty upon property value for not being directly located on the Franklin route. Once the EmX was implemented, not only was there an increase in the effect of being located within the corridor, but there was also an increase in the penalty of not being located directly on the route, confirming the hypotheses that EmX had a positive effect on property values in the corridor and that this effect was distance dependent. The base and trend model was then used to formulate a separate, dual-model that compared trends along the West 11 th and Franklin corridors; the results are detailed in Table 2. This new model was then used to forecast potential changes in the value of commercial and mixed-use properties along the West 11 th corridor that may be realized due to the implementation of the proposed EmX extension. It was found that property values have been decreasing in time over the past twelve years due to characteristics of the West 11 th corridor unrelated to LTD and EmX. Though the model predicts the trend

22 in property values to remain negative once the proposed extension is implemented, it appears this negative trend will be smaller in magnitude than it would be if the proposed EmX extension is not implemented. The accuracy of the models is questionable due to omitted variable biases stemming from the lack of sketch information for the commercial properties in the RLID data set. Specifically, the fact that both the omitted sketch variables and the franklin_test variable are positively correlated with the value of a given property implies the coefficient associated with franklin_test is likely to be overestimated. Moreover, the coefficient associated with total_acreage is also likely to be overestimated for the same reason. That being said, all results--though not necessarily the magnitude of the results-- were consistent with the hypotheses of this paper. That is, there exists an increase in the value of commercial and mixed-use properties due to the implementation of the EmX, a result that could prove very consequential to the future of the EmX system in Eugene and potentially to the bus rapid transit industry nationwide.

23 VI. Works Cited 1. Vincent W, Callaghan Jerram L Bus Rapid Transit and Transit Oriented Development: Case Studies on Transit Oriented Development Around Bus Rapid Transit Systems in North America and Australia Breakthrough Technologies Institute. Washington, DC. 2. Lane Transit District. EmX History. Web. 1 Mar < c741bf4ac81e3>. 3. Pittman, A Green Dragon Bus Eugene Weekly. Eugene, OR. 4. Lane Transit District. Fact Sheet: West Eugene EmX Extension. Web. 1 Mar < LTD-Oct2011.pd> 5. Daily Emerald Archies West Eugene Parkway Stopped For Now. Daily Emerald. Eugene, OR. 6. Perk V, Catala M Land Use Impacts of Bus Rapid Transit: Effects of BRT Station Proximity on Property Values along the Pittsburgh Martin Luther King Jr. East Busway. National Bus Rapid Transit Institute. Tampa, FL. 7. Landis L, Guhathakurta S, Huang W, Zhang M Rail Transit Investments, Real Estate Values, and Land Use Changes: A Comparative Analysis of Five California Rail Transit Systems. University of California, Berkeley. Berkeley, CA. 8. NEORail II The effect of rail transit on property values: A summary of studies. NEORail. Cleveland, OH. 9. Weinberger R Commercial Property Value and Proximity to Light Rail: A Hedonic Price Application. University of California, Berkeley. Berkeley, CA. 10. Hodel P, Ickler M The Value of Bus Rapid Transit: Hedonic Price Analysis of The EmX in Eugene, Oregon. University of Oregon. Eugene, OR. 11. Rosen, S Hedonic Price and Implicit Markets: Product Differentiation in Pure Competition. University of Rochester and Harvard University. Rochester, NY. Cambridge, MA.

24 VII. Figures Figure 1: Current Franklin EmX route Figure 1 details the locations of stations along the current Franklin EmX route. Figure 2: Proposed West 11 th EmX extension Figure 2 details the currently proposed West 11 th EmX extension and associated stations. It will start from Downtown Eugene Station, travel along West 6 th, a portion of Chambers street, West 11 th, and terminates towards the intersection of Beltline Pape Highway and West 11 th.

25 Figure 3: Distribution of RLID sale data Figure 3 is a histogram that illustrates the distribution of sale prices in the RLID set beginning the year 2000 through Notice the distribution is positively skewed, as one would expect when viewing property sale data. Though this observation is often consistent with heteroskedasticity, a simple Goldfeld-Quandt test revealed no significant signs of heteroskedasticity.

26 Figure 4: Map Detailing Eugene Points of Interest Blue blips on Figure 4 indicate the locations of intersections and amenities used as regressors for property value. Intersections and amenities in no particular order as follows: Autzen Stadium, University of Oregon, Lane Community College, Valley River Center, Randy Pape Beltline Highway, Randy Pape Beltline and Barger, Randy Pape Beltline and West 11 th, Randy Pape Beltline and Royal, Chamber and 18 th, Willamette and Broadway, Coburg and Randy Pape Beltline, Coburg and Cal Young, Sacred Heart, Riverroad and Randy Pape Beltline, Willamette and 29 th, and Willamette and 40 th.

27 VIII. Tables Table1: Regressions (1) (2) VARIABLES Base and Trend Model Hedonic Model ln_acres total_acres 0.549*** (0.0130) *** ( ) half_mile 0.731*** 0.362* (0.123) (0.217) franklin_test 0.415** 0.414* (0.182) (0.247) trend 0.168*** 0.142*** (0.0241) (0.0270) trend_squared *** * ( ) ( ) dist_corr_premx dist_corr_emx autzen ** (0.364) *** (0.391) *** (0.290) emuuofo 2.346*** (0.393) lcc 0.394** (0.168) vrc deltaoaks (0.115) ** (0.163) beltline *** (0.0944) beltlinebarger 0.442*** (0.153) beltlineroyal *** (0.184) chambers *** (0.0769) downtownwillandbroadway (0.174) costco (0.113)

28 oakwaycenter 2.703*** (0.259) sacredheart *** (0.356) riverroadbeltline (0.0897) willamette29th (0.191) willamette40th ** (0.228) trend_corr_premx *** (0.0358) trend_corr_emx ** (0.0230) Constant 12.70*** 14.95*** (0.0611) (0.749) Observations 1,824 1,824 R- squared Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

29 Table2: Differences in Trend (1) (2) VARIABLES West 11th Franklin ln_acres 0.511*** 0.541*** (0.0126) (0.0130) trend 0.164*** 0.149*** (0.0235) (0.0235) trend_squared *** *** ( ) ( ) trend_corr_premx *** (0.0199) trend_corr_emx *** ( ) trend_corr_ *** (0.0130) Constant 12.82*** 12.82*** (0.0578) (0.0572) Observations 1,824 1,824 R- squared Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

30 Table 3: Summary of Statistics Summary of Statistics: Franklin Corridor Variable Observations Mean Standard Deviation Minimum Value Maximum Value adj_sale_price ln(sale_price) total_acreage ln(total_acreage) dist_nearest_station Summary of Statistics: Franklin Corridor Variable Observations Mean Standard Deviation Minimum Value Maximum Value adj_sale_price ln(sale_price) total_acreage ln(total_acreage) dist_nearest_station Summary of Statistics: Off the Franklin Corridor Variable Observations Mean Standard Deviation Minimum Value Maximum Value adj_sale_price ln(sale_price) total_acreage ln(total_acreage) dist_nearest_station Summary of Statistics: Franklin Corridor Variable Observations Mean Standard Deviation Minimum Value Maximum Value adj_sale_price ln(sale_price) total_acreage ln(total_acreage) dist_nearest_station Summary of Statistics: West 11 th Corridor Variable Observations Mean Standard Deviation Minimum Value Maximum Value

31 adj_sale_price ln(sale_price) total_acreage ln(total_acreage) dist_nearest_station