Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 Effect of the Bull City Connector on Land Value in Durham, North Carolina Abstract Using a hedonic price model, this paper assesses the impact of the Bull City Connector on single family housing property values in Durham, North Carolina. Residential parcels within 1 km of the 18 bus stations are included in the model as well as independent variables that describe walking distances, property characteristics, neighborhood characteristics and neighborhood amenities. The results suggest that there exists a positive association between proximity to bus stops and property values, though this benefit is not felt evenly throughout all neighborhoods. Bus station areas with low-income families experience the highest proximity effect while some of the high-income neighborhoods experience negative effects. This result might contain upward bias because the accessibility brought by bus stations is strongly correlated with the Main Street amenities effect. Therefore, both pro-downtown development in Durham and investments in transit are likely to have accrued price premiums on housings close to Main Street and the bus stops. Introduction Urban economic theory states that there exists a relationship between transit systems and housing value. As long as there is a limited amount of land near transit stations, theory holds that those who want to live or work with accessible transit will bid up land prices. From past research results, it is generally agreed that property near heavy and commuter rails accrue the greatest gains in value; however, there are inconsistencies among the finding (Wardrip). The difficulty in reaching a consensus regarding transport and land value connection can be attributed to the complexity involved in urban development and the distinct urban environments and transport systems. The impact of transit on housing also depends on a variety of mediating factors, ranging from the housing type, housing market strength, neighborhood characteristics, and the quality and extent of the transit system (Giuliano). The Bull City Connector, launched on August 16, 2010, is a fare-free bus service in Durham, North Carolina, designed to link together key city locations, including Duke University, downtown Durham, Ninth Street and Golden Belt. This bus line is part of the city’s effort to expand its public Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 transit service to increase community mobility, access, and connectivity to further contribute to the revitalization and economic development of downtown Durham. The service is jointly funded by a federal grant, a state grant, Duke University and the City of Durham, with an additional agreement in which Duke University agreed to contribute toward Bull City Connector’s annual operating costs (“The Bull City Connector”). This paper explores whether proximity to the Bull City Connector has an impact on singlefamily property values in Durham. Although research has been done on the effect of light and commuter rail on the real estate market, similar research is almost nonexistent for bus services. It has been suggested that “proximity to bus stations only have modest gains, if any at all, because most bus routes lack the permanence of fixed infrastructure” (Hess and Almeida 2007). The Bull City Connector, however, is distinct from conventional bus services in that it is a fare-free line and is one of the most frequent and predictable bus services offered in the Durham area that connects key destinations in Durham. It is an example of the city’s transit oriented development that aims to assist in the effort of revitalization and development of downtown Durham. Thus, documenting the impact of transit provision on Durham’s economic development, jobs, housing, and land markets allows lessons to be learned and applied as more transit systems are planned for the future. Real estate prices can be used as one indicator of the degree to which transit investments confer benefits. Starting with a brief review of the past work on land value impact of transit, this paper then discusses the methodology and data sources used in this research. Next, descriptive statistics and research results are presented. Lastly, findings and their implications are summarized. Literature Review Traditional urban economic theory supports the hypothesis that a property near public transit should command a higher price than one that is located far away. This is because public transit should facilitate commuting; thus, households with easy access spend less money and time on transportation and more on housing (Lewis-Workman and Brod). However, there are also negative externalities associated with living too close to transit, such as noise or pollution (Chen, Rufolo, and Dueker). Most, though not all, studies on the impact of transit on single-family housing value have found property premiums for units closer to stations. The “hedonic pricing model” is often used to Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 isolate the effect of location from other property characteristics that might affect property values. In addition, although there is a consensus that locating closer to transit adds premium to a property’s value, the magnitude of the added premium is more difficult to agree on. The largest premiums appear to come from heavy rail lines and commuter rails in big metropolitan areas where public transit brings significant accessibility benefits (Landis et al.; LewisWorkman and Brod). A transit system that has geographic coverage through an area with important destinations (job centers, commercial hubs, restaurants) commands a greater increase on housing value than one running through an area with fewer amenities. Frequency, speed, and scope of service also influence the magnitude of impact. Previous studies have also indicated that impact of transit systems varies across different station areas. The most significant distinction is made between downtown areas and all others (Cervero and Landis). For example, in San Diego, using hedonic price models, r, t 10 percent to 17 percent premiums from being within a half-mile distance ring of the stations of San Diego Trolley Line (Cervero). The Hiawatha LTR line in Pittsburgh has generated $18,374,284 worth of housing premium for single-family homes, and $6,900,598 for multifamily homes (Goetz et al.). In Buffalo, a typical home located within one quarter of a mile of a transit station can earn a premium of $1300 – 3000, which is 2 to 5 percent of the city’s medium home value (Hess and Almeida). There are also a few studies that find no effect or even a negative effect of the proximity to transit. This is attributed to nuisance effects caused by locating too close to certain types of transit stations, as well as the lack of significant accessibility improvement brought by the additional transit service. People might also drive to stations, further reducing the benefits of proximity to transit nodes. For example, light rail system in San Jose and Sacramento, the CalTrain is shown to have no impact on housing prices (Landis et al.). Similar studies on bus services is much rarer. Barker, in his “Bus Service and Real Estate Values”, suggests that while the exact relationship between distance to bus stops and property values are not clear, there usually exists a positive correlation between the two (Barker). Accessibility to transit is generally measured in terms of distance to the nearest transit stations. This can be either as a measurement of straight-line distance (also known as the perceived distance) or the network distance that a pedestrian has to travel in order to get from a property to Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 the closest station. A common research method is to set up a buffer area from ¼ miles to 1 mile around stations, the acceptable walking range to transit stations, as a proxy for time spent traveling. Hypothesis The Bull City Connector is distinct from other bus services in that it is intended to offer “a different kind of transit experience – one which attracts ‘riders of choice’”. The targeted population is not only those forced into public transit by necessity, but also includes those who voluntarily choose bus over private cars because of its convenience. The Connector has a high frequency of service, running every 17 minutes between the hours of 6:30 am and 9 pm weekdays and 10 am and 9 pm Saturdays. Its route links key destinations in Durham, and its fare-free service adds to its advantage over traditional buses. Therefore, even though previous literature indicate an insignificant relationship between proximity to bus stop and property value, my hypothesis is that there exists a positive and significant correlation between the two. Methodology Study Area Debuting on Aug. 16, 2010, the Bull City Connector is a bus service that runs from Duke University campus, across Durham Downtown through Ninth Street, to Golden Belt. It operates six days a week, makes 18 stops on each way. However, the bus route has changed since Aug. 15, 2015 due to central Durham expansion. Detours off Main Street to Durham Station and Durham Center are cancelled to straighten out the route, cutting travel time between West and East Durham. The route also has been extended, continuing West on Erwin Road to a terminus on Research Drive on Duke University’s West Campus (see Map 1 and Map 2). Because digital data on the location of previous bus stops can no longer be accessed, the bus stops analyzed in this paper are from the new bus route. This change in route should not have a significant impact on the results because only two short detours have been removed. Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 Past literature has indicated that 400 meters to 800 meters are acceptable walking distance to transportation facilities (Untermann 1984). This paper chooses a slightly wider radius around the stations at 1000m to capture the variation in values not necessarily observed in the 800m zone. Single-family housings within this radius are included in the analysis. The typology of station areas for the Bull City Connector ranges from downtown urban nodes to residential stations. Given that surrounding land uses and neighborhood characteristics may have impacts on the economic development around the stations, Table 1 shows a summary for land uses around each station according to parcel data collected in 2010. Stations are arranged beginning from the western terminus at the Duke Research Lab and ending at the eastern terminus of Durham Hosiery Mills (Golden Belt with the westbound bus). Because westbound and eastbound bus stops are in close proximity to each other, the analysis is based on the eastbound bus stops. As shown in Table 1, only seven out of the 15 bus stops have a relatively significant percentage of single-family housing parcels. Therefore, an individual regression is developed for the seven stations with the highest percentage of residence housings and one aggregate regression model is developed to encompass all of the stations. Map 1: The Bull City Connector Old Bus Route (Before August 2015) Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 Map 2: The Bull City Connector New Bus Route (After August, 2015) Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 Table 1. Land Use Statistics by Bull City Connector Station Area Station Area # of Total # of Single Family % of Single Family Parcels Residential Housing Residential Housing Duke Research Labs 195 87 0.446 North Pavilion 502 259 0.516 Duke West Campus 717 454 0.633 Erwin Square/ Duke 1073 620 0.578 Ninth St. District 1211 724 0.598 Duke East Campus 1304 719 0.551 Buchanan 1269 636 0.501 1332 573 0.430 Amtrak/West Village 1270 431 0.339 Five Points 1161 325 0.280 City Center/CCB 1157 333 0.288 Mangum St 1216 372 0.306 Durham County 1396 524 0.375 Health Department 1512 607 0.401 Durham Hosiery 1611 805 0.500 Central Campus Blvd/Brightleaf (West) Brightleaf Square (Main) Plaza Government Mills Data Sources Datasets used in this research include the parcel data provided by Land Record/GIS Department from the Durham County and the census data from SimplyMap. A property tax map Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 with attributes, the parcel data are maintained by Durham County Tax Administration in order to perform equitable tax assessment and collection in Durham County, North Carolina. The map provides appraised values of all parcels located in the county, as well as information on the parcel’s area, structure size, structure age and property ownership changes. The Tax Administration data are joined with the geographic information systems and the census data obtained from SimplyMap. SimplyMap is a web-based mapping and data analysis application that enables users to develop thematic maps and reports. Its database contains nation-wide demographic, business and marketing variables on census block-groups, census tracts, ZIP codes, cities, counties, states, and the United States. Data on the site are provided by Easy Analytic Software Inc., Applied Geographic Solutions, Inc., Mediamark Research, Inc., and D&B. This paper obtains from Simply Map its data on median household income and percentage of renter occupied housing, both on a block group geographical level. Variables related to distance, such as proximity to bus stops and local amenities were calculated using Geographic Information System (GIS) tools. Hedonic Price Modeling A hedonic pricing model is used in this paper to separate out the effects of housing attributes and local amenities to determine the marginal impact of distance to a bus stop on property value. This model assumes that the price of a property is comprised of a bundle of attributes and amenities. To estimate the influence of the land’s location on its price, the following hedonic regression model is used: P = f (D, C, N, L) P equals the assessed value of the parcel, which is a function of four vectors of independent variables; D is a vector of one variable that measures walking distances from parcels to the closest bus station; vector C is comprised of property characteristics; N is a vector of neighborhood characteristics; and L describes the location amenities. Table 1 lists the nine independent variables taken into account in this hedonic price model, along with the units that they are measured in and the source that the data comes from. Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 The challenge in measuring the influence of proximity on property value is to parcel out this effect vis-à-vis all of the other factors that influence prices. Apart from structural characteristics such as the age and area of a parcel, its neighborhood characteristics and regional amenities are also taken into consideration. American Tobacco campus serves as a proxy for downtown Durham in this paper because of its central location and the cultural and commercial activities regularly hosted there. Because the Bull City Connector operates along Main Street, a major commercial street that runs through the heart of downtown, it is also important to control for the Main Street amenities effect. Table 2. Vectors, Variables, Units and Data Sources Vector Variable(s) Units Data Sources Price Assessed Property Value $/m2 Parcel Data Distance Walking Distance Along Meter Computed with GIS Street Network to Closest Bus Stop Property Characteristics Area of Parcel Square Meter Parcel Data Age of structure Years Parcel Data Neighborhood Medium Household $ SimplyMap Characteristics Income % SimplyMap % Housing, renter occupied Locational Amenities Straight Line Distance to Meter Computed with GIS American Tobacco Straight Line Distance to Meter Computed with GIS Durham Freeway Straight Line Distance to Meter Main Street Computed with GIS Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 Use of GIS A Durham county street map provided by Durham GIS services served as the base layer in geographic information systems (GIS). Parcel data and census data were then introduced as two separate layers and were spatially joined. Geographic analysis was performed to assign each parcel the corresponding block level census data if the centroid of the parcel lies within the block group boundary. The Bull City Connector bus stops were coded into GIS using coordinates identified through Google Maps, as shown in Figure 1. Using the Durham map layer, locational amenities were identified, which include Durham Freeway, American Tobacco Campus and Main Street, and for each amenity, separate layers were created to facilitate distance calculation performed later. The “Near” command is used for measuring the distance between each parcel and the highway, American Tobacco and Main Street. Walking distances from each parcel to the nearest bus stop is computed by first creating a network dataset from the Durham GIS street map. Bus stops are loaded as closest facilities and parcels are imported as incidents. Closest facility analysis solved the distance from each parcel to the nearest bus stop along the street network. Out of the 2876 parcels within one-kilometer buffer zone, the distance from 2607 parcels to bus stop were successfully solved. The other 269 incidents yielded errors because of defects in the road network. The streets on which the unresolved parcels are located are not properly joined to the road network, and thus, no continuous route can be found by GIS. They were discarded because 269 is only a small portion of the entire dataset. Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 Figure 1: GIS Interface of the Bull City Connector Bus Route. Parcels that are highlighted blue are single-family homes within one-kilometer buffer zone of bus stations. The red bolded line indicates the Durham Freeway. Analysis This paper looks at the approximate walking distance from parcels to bus stops to investigate transit’s impact on real estate market. All predictor variables are first visually examined with scatter grams in Stata. Data points that significantly diverge from the median values by visual examination are dropped as outliers and functions are transformed to better fit the normal distribution. There are 2403 parcels included in the final regression model. Two interaction variables are added, one between distance to bus stops and distance to Main Street in an effort to isolate the Main Street effect on housing prices, and the other between age of housing and the walking distance to bus stops to investigate whether there are any relationships between these variables. The all-stations model gives a summation of relationships between proximity to transit nodes and property values along the bus route. Individual regressions done on particular residential station areas indicates whether the same effect is distributed equally throughout the system or if it is more accentuated in certain residential nodes. Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 Descriptive Statistics Table 3 provides the descriptive statistics for the independent variables in this hedonic regression model, including their mean, maximum, minimum and standard deviation. There are 2876 single-family residential parcels within one kilometer of the bus stops, but only 2403 parcels are analyzed. The rest of the parcels are either located on city streets that are not properly joined to the road network (no walking distance can be found), or the parcel information contains outliers and are dropped during visual examination. Table 3. Descriptive Statistics of Variables Used in Hedonic Price Model Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 Variable Name Description Minimum Maximum Mean Standard Deviation AREAM2V Assessed value of the 24.63944 799.7922 162.3723 103.5814 157.3937 3315.202 987.0028 312.1783 98.18669 3683.932 721.9454 298.5382 32 76 56.15148 12.5783 17740 100000 46762.49 20864.19 66.08 99.11 78.37209 7.470084 19.64345 1602.015 714.7034 411.4413 162.0689 4386.611 1773.929 803.1914 15.07722 1835.321 603.8939 279.6836 single-family residence (land + structure) per square meter DIST Walking distance (in meter) of parcel to the nearest bus station AREA Size of parcel in square meter AGE Block level data for the median age of structures MHINCOME Median household income for block group within which the parcel is located PRENT % renter occupied housing for block group within which the parcel is located DHIGHWAY Linear distance in meter from parcel to the Durham freeway DAMERTOB Linear distance in meter from parcel to American Tobacco DMAIN Linear distance in meter from parcel to Main Street Findings Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 All Stations Model Variables are included in this hedonic model if past research indicates that they are important components of housing values. Table 4 shows a summation of price-distance relationships observed at all of the stations along the Bull City Connector. The R squared value indicates that this model explains close to 42 percent of the variation in residential property price. This R squared value, although lower than those of previous literature, is expected because of many key variables usually included in hedonic models are unavailable for the housing stock in Durham, such as individual structural age or the structure area as opposed to land area. All of the predictor variables are significant at the 1% probability level. Model results suggest that the distance of a parcel to a station and its property value are inversely related – housing prices decreases as distance increases. Decreasing marginal effects are also found. For example, for houses that are 56 years of age (the median age for parcels analyzed in this paper) and 100m in linear distance from the Main Street, the per meter squared property value of a single-family home decreases by approximately 18 percent when the walking distance increases from 200 to 300 meters away from a bus stop. Houses located at a distance of 500 meters from Main Street, also aged at 56 years, the per meter squared property value decreases only approximately 3.4 percent when the walking distance increases from 900 to 1000m. Figure 2 shows the relationship between walking distances and the percentage change in housing values for houses at the median age of the dataset. This result is relatively large when compared to findings of past literature on the impact of proximity to transit stations. This upward bias in result is likely due to the difficulty in differentiating benefits accrued from amenities near the stations and the actual station effect. Although two interaction variables are introduced to the model in an effort to correct for this spatial autocorrelation, there might still be some unobserved relationships between the variables that are assumed to be independent, skewing the result. This result, however, does clearly suggest a high centrality to Durham – Durham residents value living close to Main Street and the Bull City Connector because of the accessibility benefits that the location brings. Relationships between other attributes and property value can also be drawn from Table 4. Consistent with previous findings (Gamble et al.), houses closer to highways depreciate in value likely due to disamenities such as noise and air pollution created by proximity to highways. In this Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 case, parcels near the Durham Freeway has lower property value than those far away. This negative relationship might also be attributed to historical reasons because the neighborhood surrounding the highway has always been an area notorious for crime and poverty until recent development. Interestingly, proximity to American Tobacco Campus seems to have negative impact on housing value. This result is unexpected because American Tobacco Campus encompasses restaurants, Durham Performing Arts Center, Durham Bulls Athletic Parks and other venues for entertainment events and commercial activities in Durham. These amenities should bring higher values to homes located nearby. One plausible explanation for this home value discount is that American Tobacco Campus is situated in a mainly commercial neighborhood with few residential housings. The residential neighborhood directly east of this district, however, has been a historically low-income neighborhood. Therefore, in this case, the neighborhood effect dominates over the accessibility benefits. For variables associated with neighborhood characteristics, the results show that the value of single-family housings increases with neighborhood median family income, which is congruent with expectation. In assessing the effect of property age on property values, according to the model results, there is a positive correlation between the two until a house reaches 50 to 60 years of age. Then, the positive effect begins to trail off and becomes slightly negative. The independent variable of percentage of renter occupied housing in a neighborhood is affected by a combination of its linear, square, and cubic functions, all terms significant at 99.99% level. When the effect is plotted out, Figure 3 indicates that the housing value rises until the percentage of renter occupied housing reaches 70 to 80 percent. From there, the price falls at a decreasing rate. Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 Table 4: Regression Model Results for All Stations Single Family Housings Variables Coef. Std. Err. t P>t [95% Conf. Interval] lnDIST -0.4302405 0.17912 -2.40 0.016 -0.7814872 -0.0789938 AREA -0.0005594 0.0000308 -18.16 0 -0.0006198 -0.0004989 AGE 0.1026589 0.0143354 7.16 0 0.0745477 0.1307701 sqrAGE -0.0003814 0.0000772 -4.94 0 -0.0005328 -0.00023 sqrtMHINCOME 0.0068364 0.000218 31.36 0 0.0064089 0.0072639 PRENT 3.564968 0.4353134 8.19 0 2.711337 4.418599 sqrPRENT -0.0419341 0.0054123 -7.75 0 -0.0525473 -0.031321 cubePRENT 0.000162 0.0000222 7.28 0 0.0001183 0.0002056 lnDHIGHWAY 0.1085767 0.0139855 7.76 0 0.0811517 0.1360016 lnDAMERTOB 0.1574145 0.0261187 6.03 0 0.1061968 0.2086323 lnDMAIN -0.6747746 0.1987478 -3.40 0.001 -1.064511 -0.2850386 InmainmINTl~d 0.1152055 0.029159 3.95 0 0.0580259 0.1723851 ageINTlnwal~e -0.0106868 0.0018261 -5.85 0 -0.0142677 -0.007106 Summary Statistics: Number of Observations = 2403 R-Squared = 0.4195 Adj R-Squared = 0.4163 Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 Figure 2. Relationship between Proximity to Bus Stations and Housing Prices 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 0.1 0 -0.1 -0.2 -0.3 -0.4 -0.5 -0.6 -0.7 -0.8 100m to Main St. 500m to Main St. 1000m to Main St. 2000m to Main St. Figure 3. Relationship Between Structure Age and Property Value 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 3000 Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 Figure 4. Relationship between Percentage of Renter Occupied Housing and Property Value 65 67 69 71 73 75 77 79 81 83 85 87 89 91 93 95 97 99 Individual Model Table 5 shows the result of individual regressions done on the seven neighborhoods that have the highest percentage of residential housings. Listed is the number of observations, median household income level, and the impact of proximity to bus stations in each neighborhood. It can be derived from the table that benefits of locating close to bus stations are not felt equally throughout the bus route. Six out of the seven station areas are relatively affluent neighborhoods whose average median income are higher than the mean income level of all the parcels examined in this paper ($46762). Only two of these six neighborhoods have shown a positive relationship between proximity to bus stops and property values. For the rest, locating closer to bus stops actually depreciates the price. In contrast, the one lower-income neighborhood – Durham Hosiery Hills, shows a positive correlation between shorter distance to bus stop and increased housing value. This finding confirms the assumption that low-income families are willing to pay a higher premium to live closer to a transit node because they lack alternatives for other modes of transportation. Higher income families who can afford automobiles are not likely to value the fare free bus service provided Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 by the Bull City Connector, particularly when the bus route is relatively short, only running across downtown Durham. Table 5. Regression Results for Individual Station Single Family Housings Station Number of Observation s Median Household Income North Pavilion 263 63394.39 Positive(+)/Negative(-) Relationship Between Proximity to Closest Bus Station and Property Value + Duke West Campus 440 63887.94 + 590 64691.47 - Ninth St. District 592 66118.18 - Duke East Campus 484 61009.52 - Buchanan Blvd/Brightleaf (West) 523 61988.98 - Durham Hosiery Mills 803 27852.84 + Erwin Square/ Duke Central Campus Conclusion This paper confirms that there exists a positive relationship between proximity to the Bull City Connector bus stations and property values in Durham on an aggregate level. Despite the lack of infrastructure for bus services when compared with light rail or commuter rails, evidence is found that in Durham, locations closer to stations accrue sufficient accessibility benefits that are reflected in the single-family real estate market, though the relationship varies considerably by station areas and neighborhood income level. The biggest premiums are accrued in lower income neighborhoods. Few higher income neighborhoods exhibit a similar positive correlation between distance to bus stop and property value, with some experiencing even property value discounts. Note that this research does not indicate any causation relationship between transit investment and property value changes in Durham because no data on the before and after bus service home sale prices is examined. The hedonic model does suggest an association between higher property value and proximity to bus stops on an aggregate level. It also suggests evidence of Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 mono-centricity in Durham because bus stations are located on one of the most commercialized streets that runs across downtown Durham. Also confirmed are some findings in past literature. Proximity to highway shows disutility effects that depreciate the value of homes close to Durham Freeway. The assumption that public transit is valued more by lower-income families is also implicated. References Barker, William G. “BUS SERVICE AND REAL ESTATE VALUES.” N.p., 1998. trid.trb.org. Web. 2 Nov. 2015. Cervero, Robert. “Effects of Light and Commuter Rail Transit on Land Prices: Experiences in San Diego County.” Journal of the Transportation Research Forum 43.1 (2010): n. pag. journals.oregondigital.org. Web. 30 Nov. 2015. Cervero, Robert, and John Landis. “Twenty Years of the Bay Area Rapid Transit System: Land Use and Development Impacts.” Transportation Research Part A: Policy and Practice 31.4 (1997): 309–333. ScienceDirect. Web. Chen, Hong, Anthony Rufolo, and Kenneth Dueker. “Measuring the Impact of Light Rail Systems on SingleFamily Home Values: A Hedonic Approach with Geographic Information System Application.” Transportation Research Record: Journal of the Transportation Research Board 1617 (1998): 38–43. trrjournalonline.trb.org (Atypon). Web. Gamble, Hays B. et al. “The Influence of Highway Environmental Effects on Residential Property Values.” Penn State Univ Inst for Res on Land and Water Resour Res Publ 78 (1974): n. pag. ProQuest. Web. 1 Dec. 2015. Giuliano, Genevieve. “Public Transit as a Metropolitan Growth and Development Strategy.” Washington DC: Brookings Institution 3.Urban and Regional Policy and Its Effects (2010): 205–252. Print. Hess, Daniel Baldwin, and Tangerine Maria Almeida. “Impact of Proximity to Light Rail Rapid Transit on Station-Area Property Values in Buffalo, New York.” Urban Studies 44.5-6 (2007): 1041–1068. Print. Landis, John et al. “Rail Transit Investments, Real Estate Values, and Land Use Change: A Comparative Analysis of Five California Rail Transit Systems.” eScholarship (1995): n. pag. escholarship.org. Web. 30 Nov. 2015. Lewis-Workman, Steven, and Daniel Brod. “Measuring the Neighborhood Benefits of Rail Transit Accessibility.” Transportation research record. 1576 (1997): n. pag. Print. Angela Chen Professor. Charles Becker Urban Economics November. 30th, 2015 “The Bull City Connector.” Let’s Go Durham-Orange. N.p., n.d. Web. 30 Nov. 2015. Wardrip, Keith. Public Transit’s Impact on Housing Costs: A Review of the Literature. Center for Housing Policy, 2011. Web. 30 Nov. 2015. Insights.