Quantifying Transit’s Impact on GHG Emissions and Energy Use—The Land Use Component (2015)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

13 S E C T I O N 4 4.1 Summary of Key Findings There are two key aspects of the land use effect: 1. The effect of existing transit systems and 2. The effect of current or potential future transit system expansions or enhancements. The research team used slightly different methods for analyzing each effect. 4.1.1 Effect of Existing Transit Systems The effect of existing transit systems is best examined at the regional level, in order to capture the entire transportation and land use ecosystem, as described above. Each urban region of the United States has had many years to arrive at a relative equilibrium of transportation and land use, despite some ongoing marginal changes. In particular, large, older cities on the East Coast and in the Midwest and some West Coast cities like San Francisco and Los Angeles have rich histories of development around transit infrastructure. The effect of the existing transit system measures the cumulative effect of that entire history to the present day. To describe the land use effect of existing transit systems in a different way, consider the dif- ference between a city with a compact core and a historically robust transit system, such as New York, and a city with little distinct core and far less transit, such as Dallas. The regional population density of New York is 4,176 people per square mile, and average daily per capita VMT is 15.8. The regional population density of Dallas is 2,149 per square mile, and average daily per capita VMT is 24.2. Without its dense transit network, New York may have developed more like Dallas, with lower population densities and a more car-dependent transportation system. Of course, transit is not the only factor that shapes land use and travel patterns. Other factors include geography and economic and technological variables. The statistical analysis in this research calculates the share of the “compactness” of a given region that can be attributed to transit: the land use effect of transit. Key findings about the land use effect of existing transit systems are as follows: • Effect on population densities. Taking the entire U.S. urban population in aggregate, gross population densities would be lower by 27% without transit systems to support compact development. In other words, U.S. cities would consume 37% more land area in order to house their current populations. The land use effect of existing transit makes U.S. cities more compact. • Effect on VMT, fuel use, and transportation GHG. By providing more walking and biking opportunities and making some journeys by car shorter, the land use effect of transit produces an aggregate 8% decrease in VMT, transportation fuel use, and transportation GHG emis- sions in U.S. cities. The Land Use Effect of Transit: Findings

14 Quantifying Transit’s Impact on GHG Emissions and Energy Use—The Land Use Component • Effect of transit trips replacing automobile trips. By transporting people on buses and trains that would otherwise travel by automobile, transit systems also produce a complementary ridership effect. In aggregate across U.S. cities, transit ridership reduces VMT, transportation fuel use, and transportation GHG emissions by 2%. This is a substantial change given that only 4% of passenger trips are currently made by transit in U.S. metropolitan areas. • The land use benefit of transit varies across urban areas, ranging from a 1% to 21% reduction in VMT, transportation fuel use, and transportation GHG emissions compared to a hypo- thetical scenario without transit. Urban areas with higher route densities of transit, service frequencies of transit, and availability of light rail have higher land use effects. Not surpris- ingly, higher land use effects of transit are generally found in more densely developed areas. • The land use effect of transit in a given region typically reduces GHG emissions more than the ridership effect. The average ratio of land use benefits to ridership benefits across all U.S. cities is 4:1, but the ratio varies substantially across different urban areas.2 The statistical models developed in this research find that roads have the opposite effect on land use. Generally speaking, transit competes with the private automobile as a mode of per- sonal transportation. There is a discernible tradeoff between investing in roads and investing in transit, and this tradeoff extends to the land use effect. Travel by private automobile consumes more space than travel by transit, with drivers requiring both roadway and parking space for their vehicles. 4.1.2 Effect of Current or Potential Future Transit System Expansions or Enhancements The marginal effect of transit system expansions or enhancements is measured at a different scale. These include expansions of individual or multiple routes, enhancements to transit level of service on individual or multiple routes, or additions of new transit modes. Each of these improvements has the ability to incrementally increase the land use effect of transit over time. The marginal effect measures the change in land use patterns and associated travel patterns that are attributable to the improvement. Since land development is a relatively slow process, with even proactively planned developments sometimes taking more than a decade from planning to occupancy, it can take many years to realize the land use effect of new investments. Key findings about the land use effect of system expansions or enhancements at the regional level are as follows: • Increasing transit route densities (route miles/land area) by 1% in a region is associated with an increase in population density of 0.2%. The corresponding land use benefit is a 0.05% reduction in VMT, transportation fuel use, and transportation GHG emissions. • Increasing transit service frequencies by 1% in a region has nearly the same effect: an increase in population density of 0.2%. The corresponding land use benefit is a 0.04% reduction in VMT, transportation fuel use, and transportation GHG emissions. Key findings about the land use effect of system expansions or enhancements at the neighbor- hood level are as follows: • Adding a rail station to a neighborhood that did not previously have rail access is associated with a 9% increase in activity density (combined population and employment density) within a 1-mile radius of the rail station. The corresponding land use benefit is a 2% reduction in VMT (for households within the 1-mile radius), transportation fuel use, and transportation GHG emissions. 2 Complementary ridership effects of transit vary based solely on the level of transit ridership in individual regions.

The Land Use Effect of Transit: Findings 15 • Improving employment accessibility by clustering new jobs around transit nodes or improv- ing the bus and rail network in individual neighborhoods can also have potent land use effects (described in more detail in Section 4.4.2). • An analysis of the Portland Westside light-rail extension found a land use effect of 24% increase in densities in the corridor area between 1994 and 2011. These changes correspond to a 6% household VMT reduction due to the land use effect and an additional 8% VMT reduction due to the ridership effect. 4.2 How to Measure Density? In order to study the land use effect, what constitutes compact development and how it is measured must be clarified. Typical characteristics of compact development versus sprawling development are higher densities, more land use mixing, better access to transit, a more pedes- trian-friendly environment, and closer access to regional destinations, especially jobs. These characteristics in particular are the ones associated with lower VMT. Density is the most commonly referenced and most easily measured indicator of compact development. Density is commonly measured in terms of population and/or jobs per square mile. But density can be characterized at different geographical scales. Both local and regional densities matter to travel patterns. Local densities are easily observed—development patterns are clearly denser in Rosslyn, Virginia, than in Fairfax, Virginia. At the regional scale, density is more challenging to characterize, as metropolitan regions are made up of numerous cities and neighborhoods that can vary widely in development style. At the regional scale, gross density is the easiest to measure, dividing total regional population by total regional land area. Gross regional density is a reasonable measure of density for the pur- poses of this research because higher gross densities are associated with lower per capita VMT (as discussed in the following section), but gross densities also mask important subregional variations. The New York City and Los Angeles Metropolitan Statistical Areas have very similar gross population densities at 2,826 and 2,646 people per square mile, respectively (U.S. Census Bureau 2012). But the New York City region has a super dense core with sprawling suburbs. The Los Angeles region has little distinct core, but moderate uniform density throughout. In New York, many people are living at much higher local densities than almost anyone in Los Angeles. Population-weighted density is an emerging alternative way to measure regional densities accounting for local variations. Densities are first calculated at the local scale, for example popu- lation per square mile in each census tract. Regional density is then calculated as the average of local densities, with each census tract’s density weighted by its population. In this way, census tracts where more people live (which tend to be more densely populated tracts), are given more weight in the calculation. Population-weighted density is a better regional measure of the typical local density experience of residents. The population-weighted density of New York at 31,251 people per square mile compares with that of Los Angeles at 12,114 (U.S. Census Bureau 2012). In this research, gross density is used as the measure of regional density because gross density is readily measurable with available data, whereas population-weighted densities are extremely time intensive to calculate for multiple custom geographies.3 Gross density is also a reasonable predictor of travel patterns and has been used extensively in the literature on the topic. However, it should be kept in mind that gross density is a relatively simple proxy measure to describe com- plex variations in urban form. The land use effect of transit can contribute to changes in urban form that are not fully captured by gross density. Using gross density in statistical models could 3 Regions were defined by FHWA boundaries for metropolitan areas.

16 Quantifying Transit’s Impact on GHG Emissions and Energy Use—The Land Use Component understate the magnitude of the land use effect. Still, using gross density to analyze the land use effect provides a solid link between transit systems and travel patterns. 4.3 Land Use Benefits of Existing Transit Systems Transit systems in every U.S. city have a land use effect, and these effects vary in magnitude based on the density and quality of transit service. The research team estimated the strength of the effects between key variables in order to construct a model of the transportation and land use ecosystem. By manipulating inputs to the model, the size of the land use effect is estimated in two stages. First, effects of transit on land use (the land use effect) are estimated. Second, effects of land use on VMT, fuel consumption, and GHG emissions (land use benefits) are estimated. 4 The elasticity of VMT with respect to density of -0.3 is based on the findings of Ewing and Cervero in “Travel and the Built Environment: A Meta-Analysis” (2010). While the models constructed in this study suggest lower elasticities, these represent only the relationship of density to travel patterns. Other key “D” variables, including Diversity (land use mixing), Design (pedestrian environment), and Destinations (regional accessibility) are not included in the model. Given that denser places usually score higher on the other “D” variables as well, it is appropriate to adjust the elasticity upward to account for these missing variables. 5 Bureau of Transportation Statistics. Table 1-6: Estimated U.S. Roadway Lane-Miles by Functional System. Office of the Assis- tant Secretary for Research and Technology, U.S. DOT. http://apps.bts.gov/publications/national_transportation_statistics/ html/table_01_06.html. 6 APTA. 2012 Public Transportation Fact Book. Washington, D.C., September 2012. http://www.apta.com/resources/statistics/ Documents/FactBook/APTA_2012_Fact%20Book.pdf. 7 2009 National Household Travel Survey (NHTS) Includes all buses, trains, streetcars, trolleys, and ferries. Excludes taxicabs. The effect of existing transit systems is measured using linear structural equation modeling (SEM) based on data from a sample of over 300 urbanized areas. The trans- portation, demographic, and land use data used are from 2010. Complete technical details of the model are provided in “Appendix B: Statistical Models in Depth.” 4.3.1 National Land Use Benefits Taking the entire U.S. urban population in aggregate, gross population densities would be lower by 27% without transit systems to support compact development. In other words, U.S. cities would consume 37% more land area in order to house their current populations. That is a dramatic difference in urban character, with direct implications for travel patterns, energy use, and GHG emissions. Higher densities bring destinations closer together, allowing for shorter car trips and more walking, bicycling, and carpooling. Using the elasticity of VMT with respect to density of -0.3 (as discussed in Section 3), the U.S. population living in cities without transit would see its VMT increase by 8% due to lower population densities.4 The ridership effect, when transit riders would be forced to begin driving, would increase VMT an additional 2%, for a total VMT increase of 10% if transit were eliminated altogether. These numbers must be understood relative to the scale of investment in different transporta- tion modes. In every city in the United States, infrastructure dedicated to private vehicle travel dwarfs public transportation infrastructure. There are 8.6 million lane miles of roadways in the United States.5 In comparison, there are 244,000 directional route miles of transit service.6 Not surprisingly then, transit represents a very small proportion of total travel in the United States. Only 4% of all trips are made by transit. In contrast, 84% of trips are made by driving or riding as a passenger in a private vehicle (10.4% of trips are walking trips and 1% are made by bicycle).7

The Land Use Effect of Transit: Findings 17 Therefore, a combined 10% increase in VMT without transit (combined ridership and land use benefits) indicates the broad influence of transit systems on travel patterns. 4.3.2 Different Cities, Different Land Use Benefits Land use benefits can be estimated for individual cities using basic data on the transit system extent and level of service. In brief, cities with higher transit route densities and levels of service and cities with light-rail transit (LRT),8 have higher land use benefits. (More information about the specific data points and calculation methods are provided in “Appendix A: Key Results from Statistical Models.”) The research team calculated land use benefits individually for all 300+ cities in the urbanized areas dataset. The resulting land use benefits for the full sample of 300+ cities range from a 1% decrease to a 21% decrease in VMT. These estimates are based on gross population densities. Table 1 shows estimated land use benefits for a sample of cities. For comparison, ridership ben- efits (the additional VMT that would be created if transit riders began driving instead) estimated by the model are also shown.9 The model estimates the highest land use benefits for historic transit cities like New York and San Francisco; for newer cities, such as Portland, which have invested heavily in transit in recent years; and for some smaller cities such as Ames, Iowa, and Champaign, Illinois, that have compact cores and a relatively high level of transit service concentrated in a 8 As discussed below, light rail transit is associated with higher gross population densities. The same is not consistently true of heavy rail transit, possibly due to the potential of rail extensions into the suburbs to promote sprawl. 9 Ridership effects shown are the average of two different methods discussed in Appendices A and B of this report. Table 1. Transit land use benefits and ridership benefits for sample cities. Urbanized Area Land Use Benefit (%VMT Reduction) Ridership Total Benefit (% VMT Reduction) New York–Newark, NY-NJ-CT 19% 16% 34% San Francisco–Oakland, CA 18% 9% 27% Ames, IA 21% 4% 25% Portland, OR-WA 19% 4% 23% Champaign, IL 16% 4% 20% Washington, DC-VA-MD 12% 9% 20% Los Angeles–Long Beach, CA 15% 4% 19% Seattle, WA 14% 5% 19% Chicago, IL-IN 12% 7% 19% Salt Lake City, UT 15% 3% 18% Philadelphia, PA-NJ-DE-MD 12% 5% 17% Boston, MA-NH-RI 11% 6% 17% Eugene, OR 13% 3% 16% Sacramento, CA 13% 2% 15% Houston, TX 10% 2% 12% Austin, TX 9% 2% 11% Atlanta, GA 8% 3% 11% Kansas City, MO-KS 5% 1% 6% Greenville, SC 3% 0% 3% Note: Cities in this table were selected to represent a range of different population sizes and land use benefits. Cities are ranked from highest to lowest total benefit (combining land use and ridership benefits). Benefit (%VMT Reduction)

18 Quantifying Transit’s Impact on GHG Emissions and Energy Use—The Land Use Component relatively small urban area. The latter tend to be college towns where a high proportion of the population is made up of students, many of whom use transit regularly and do not own a car. The model estimates the lowest land use benefits for sprawling regions like Atlanta and Kansas City. The land use benefits in Table 1 quantify the reduction in driving that each region’s transit sys- tem produces by fostering compact development patterns. For example, the New York–Newark urbanized area (at 4,176 people per square mile) without its public transportation would resem- ble cities like Buffalo, New York (1,686 people per square mile), or Austin, Texas (1,750 people per square mile), in urban density. Housing the New York–Newark population at those densities would consume an additional 6,862 square miles of land. The average resident of the New York– Newark area currently drives 15.8 miles a day; without transit, residents would drive 24.1 miles a day. An additional 4.5 miles a day (19% reduction in Table 1) are attributable to the land use effect; lower densities would reduce opportunities for walking and bicycling and lengthen some car trips. An additional 3.8 miles a day (16% reduction in Table 1) are attributable to the ridership effect, as people that currently ride transit daily would increase their car travel in the absence of transit. If the Portland, Oregon, urbanized area (3,325 people per square mile) had never had public transportation, Portland would resemble a city like Ithaca, New York (1,351 people per square mile) or Fort Collins, Colorado (1,422 people per square mile) in development style. Housing the Portland population at those densities would consume an extra 788 square miles of land. The average resident of the Portland area currently drives 18.9 miles a day; without transit, residents would drive 24.5 miles a day. An additional 4.6 miles a day (19% reduction in Table 1) are attributable to the land use effect. An additional 1 mile a day (4% reduction in Table 1) is attributable to the ridership effect. It is important to keep in mind that the model results are influenced by the FHWA urbanized area boundary for each city. Estimated land use benefits vary in proportion to the density and frequency of transit within the area defined. Urbanized areas that include larger proportions of suburban development may show lower land use benefits than urbanized areas with boundaries that follow the urban core more closely, since suburban areas tend to have less transit service. Inter- ested readers can experiment with defining custom boundaries for their regional boundaries in the calculator created as a part of this research (available at www.TRB.org/main/blurbs/172110. aspx). Estimated ridership benefits vary in proportion to each area’s transit mode share. While land use benefits are typically higher than ridership benefits, there is no consistent relationship between the land use benefit and the ridership benefit across urbanized areas. For the average city, the ratio of land use benefits to ridership benefits is 4:1. For the cities listed in Table 1, ratios range from 10:1 to 1:1. Table 2 lists land use effects for the sample of cities in terms of total GHG emissions reduced. GHG emission reduction benefits are a product of the percentage VMT reduction due to the land use effect and the regional population. Larger urban areas have higher land use benefits in terms of total emissions reduced. The New York–Newark region has the highest effect of any U.S. urbanized area, with 20 billion pounds of CO2e emissions avoided due to land use benefits. Smaller cities invariably have lower total emission reductions, even if they have relatively high land use benefits in percentage terms. 4.4 Land Use Benefits of Transit System Improvements Incremental improvements to transit service have measurable incremental land use effects. Improvements include adding new bus routes or rail lines, increasing service on existing routes, and improving the overall level of access to regional employment via transit. The land use effects of improvements are measured separately at the regional level and at the neighborhood level.

The Land Use Effect of Transit: Findings 19 4.4.1 Regional Level At the regional level, land use effects of transit system improvements are measured using elasticity values derived from the urbanized area models. Increasing transit route densities by 1% in a region is associated with an increase in population density of 0.2%. The corresponding land use benefit is a 0.05% reduction in VMT, transportation fuel use, and transportation GHG emissions. Increasing transit service frequencies by 1% in a region has nearly the same effect: an increase in population density of 0.2%. The corresponding land use benefit is a 0.04% reduction in VMT, transportation fuel use, and transportation GHG emissions. These effects include both bus and rail service. Table 2. Total transit land use benefits on emissions in sample cities. Urbanized Area Land Use Benefit (%VMT Reduction) Population Land Use Benefit (Total Annual CO2e emissions reduced in lbs) New York–Newark, NY-NJ-CT 19% 18,536,839 20,045,872,992 Chicago, IL-IN 12% 8,674,561 4,407,347,990 Los Angeles–Long Beach, CA 15% 12,148,231 3,852,288,008 Washington, DC-VA-MD 12% 4,429,831 3,069,333,392 San Francisco–Oakland, CA 18% 3,334,957 2,363,357,979 Philadelphia, PA-NJ-DE-MD 12% 5,451,310 2,262,825,320 Boston, MA-NH-RI 11% 4,270,765 1,903,891,133 Atlanta, GA 8% 4,469,203 1,307,149,408 Seattle, WA 14% 3,062,739 1,209,678,011 Houston, TX 10% 4,796,260 682,165,334 Portland, OR-WA 19% 1,849,891 542,068,124 Sacramento, CA 13% 1,598,186 215,465,156 Salt Lake City, UT 15% 1,021,020 198,035,588 Austin, TX 9% 1,254,769 188,973,381 Kansas City, MO-KS 5% 1,597,839 97,779,018 Eugene, OR 13% 248,288 50,825,317 Champaign, IL 16% 143,107 35,880,621 Ames, IA 21% 59,018 10,883,718 Greenville, SC 3% 341,875 7,238,189 The effect of transit system improvements at the regional level is measured using a log SEM model based on data from a sample of over 300 urbanized areas. The transportation, demographic, and land use data used are from 2010. A more detailed description of the model is provided in “Appendix A: Key Results from Statistical Models.” Complete technical details of the model are provided in “Appendix B: Statistical Models in Depth.” For example, Los Angeles Metro’s ambitious transit expansion program can be evaluated in terms of its likely land use effects in future years. Los Angeles County is part of the Los Angeles- Long Beach urbanized area, with a gross population density of 6,251 people per square mile. The region’s transit assets include more than 900 directional route miles of rail and almost 11,000 directional route miles of bus service. Los Angeles Metro is the largest transit provider in the area.

20 Quantifying Transit’s Impact on GHG Emissions and Energy Use—The Land Use Component According to Metro’s most recent Long-Range Transportation Plan, an additional 430 new directional route miles of high-quality transit (including rail and bus rapid transit) are due to be added to the transit system by 2040. Assuming 60 vehicle trips serve each route in each direction per day and assuming average land use effects, this expansion program will lead to a 1% increase in population density in the region in the long term. The corresponding land use benefit is a reduction of regional VMT by 0.3%, saving 12 million gallons of gasoline per year and reducing GHG emissions by 116,000 tons per year. It is important to keep in mind that the effects projected here are average effects observed in existing urban areas, and that effects for individual transit system enhancements could be substantially higher or lower depending on various factors, as discussed further in Section 4.5. 4.4.2 Neighborhood Level At the neighborhood level, improvements in local transit systems and transit access gener- ally attract denser development. On average, adding a rail station to a neighborhood that did not previously have rail access is associated with a 9% increase in activity density (combined population and employment density) within a 1-mile radius of the rail station. Assuming that the location is generally suitable for rail service but does not currently have service, a neigh- borhood with 10,000 residents and 10,000 jobs could be expected to add a combined 1,800 residents and workers over time in response to a new rail station. Residents of the neighbor- hood can be expected to reduce their VMT, transportation fuel use, and transportation GHG emissions by 2% due to the land use effect, with additional reductions due to the ridership effect of transit. The effect of transit system improvements at the neighborhood level is measured using multilevel modeling (MLM) based on data from nine metropolitan regions. The date of the transportation, demographic, and land use data used varies by region. A more detailed description of the model is provided in “Appendix A: Key Results from Statistical Models.” Complete technical details of the model are provided in “Appendix B: Statistical Models in Depth.” These changes are average results expected over time. Changes around individual stations may vary substantially based on local factors. The recent experience of Evanston, Illinois, with station area developments around both existing stations and improved transit service helps to illustrate how observed changes in density relate to the model results. Evanston is a first ring suburb of Chicago. The city was originally built around transit, includ- ing streetcar and commuter rail, but had been losing population to more automobile-oriented suburbs for several decades when planning for a transit-oriented resurgence began in the 1980s. While Evanston already had five urban rail stops (served by the Chicago Transit Authority [CTA]) and two commuter rail stops (served by Metra), the city dramatically increased its support for development in station areas. The 1986 comprehensive plan called for higher density development focused around four of its most active rail stations, including zoning changes. The city also invested in sidewalk, streetscape, and utility improvements in station areas to support development. The first new downtown Evanston high rise in more than 20 years was built in 1991. Figure 2 shows the Optima Towers, built on Fountain Square in 2002, two blocks from Davis Street Station.

The Land Use Effect of Transit: Findings 21 The CTOD TOD Database (TOD for “transit-oriented development) provides several data indicators of the success of this TOD-based turnaround in terms of reversing Evanston’s overall population decline and concentrating growth around its high-capacity transit lines. Table 3 presents a summary of population and employment data from the CTOD database, for the four station areas (1⁄2-mile radius) around Evanston’s core TOD stations. These data are compared to the same data for • The station area around the Central-Metra station (not included in the city’s TOD-based growth efforts). • The station areas around the CTA-elevated and Metra stations in Wilmette, just north of Evanston. • The Chicago region. Figure 2. New development near Davis Street Station, Evanston, Illinois. Image: Flickr User Aaron Weathers Table 3. Change in activity density in Evanston station areas (1⁄2-mile radius), 2000–2010. Location Activity Density (Population and Jobs per Acre) 2000 2010 Percent Change Evanston—TOD Core Station Areas Davis 19.6 23.0 17% Dempster 11.7 13.3 13% Main 8.8 9.1 3% South Blvd 10.0 9.4 −6% Combined 12.5 13.6 9% Evanston Control Station Central-Metra 4.4 4.2 −6% Wilmette CTA 4.3 3.6 −15% Metra 4.6 4.5 −2% Combined 4.5 4.1 −8% Metropolitan Region All Areas 3.8 3.9 2% Source: CTOD TOD Database. http://toddata.cnt.org/. Jobs figures are available for 2002 and 2009 and are used as proxies for 2000 and 2010 figures.

22 Quantifying Transit’s Impact on GHG Emissions and Energy Use—The Land Use Component Areas within a half mile of the four stations increased their activity density by an average 9% over approximately 10 years.10 (When compared to the base trend of population and employ- ment growth in the Chicago region, the station areas saw a net 7% increase in activity den- sity.) This change in density is expected to lead to a 2% reduction in VMT, transportation fuel use, and transportation GHG emissions by households living in the area. Notably, the average density increase masks a wide range of variation within individual station areas, where den- sity changes over the period range from a 6% decrease to a 17% increase. Numerous factors determine the ultimate land use effect of individual transit investments, as discussed further in Section 4.5. Average changes around the core Evanston station areas from 2000 to 2010 are very similar to the average results predicted in this research of adding a new rail station to an area that did not previously have rail. Notably, Evanston’s recent experience was anchored largely by pre-existing transit service, though some new transit service was added. Improving employment accessibility can also have potent land use effects. Access to jobs and to the shopping, dining, and entertainment opportunities associated with some jobs is an important factor in residential location choice and therefore an important factor for devel- opers considering building in particular neighborhoods. The best-fit neighborhood model from this research finds that for every 1% increase in the share of regional jobs accessible by transit,11 there is an associated 0.5% increase in neighborhood activity density. The cor- responding land use benefit is a 0.1% reduction in VMT, transportation fuel use, and trans- portation GHG emissions. The importance of job accessibility is also seen in case studies of individual transit lines researched by the CTOD. An examination of development patterns around three new rail lines in Minneapolis, Denver, and Charlotte qualitatively assessed the importance of six factors in catalyzing new development around individual rail stations: proximity to downtown; proximity to employment centers; availability of vacant and under- utilized land; walkability of the neighborhood; local transit connectivity; and local household income. Proximity to employment centers was the only factor found to have a consistently strong positive relationship with development patterns around rail stations on all three lines (CTOD 2011). In Charlotte, the new LYNX Blue Line stretches 10 miles from Uptown Charlotte southward to suburban Pineville. Figure 3 provides a map of the line. Development has been strongest in the South End neighborhood, adjacent to Uptown employment centers. The South End is physi- cally cut off from Uptown by a freeway. Transit connections tapped into pent-up development demand in the South End by helping overcome this barrier, improving connections and acces- sibility between the South End and Uptown. In practical terms, transit employment accessibility can be improved in one of several ways: • Providing new transit service with connections to employment centers. The Charlotte Blue Line is an example. • Improving the speed, frequency, or connectivity of existing transit service so that employment centers can be reached more quickly. Evanston’s Davis Station area revival included improved service on the CTA Purple Line. Both bus agencies serving the station also increased their service frequencies, added stops, improved routes, and increased coordination with train schedules. 10 The neighborhood model examines changes in activity density within 1 mile of transit stations, while the CTOD database captures changes within a 1⁄2-mile radius. Thus, the comparisons provided here are not exact but are provided to illustrate general trends. If there is a 9% change expected within a 1-mile radius, and the majority of changes happen closer to the sta- tion, it is likely that changes within the 1⁄2-mile radius only are actually higher. 11 Defined as jobs accessible within 30 minutes of transit travel time from a transit stop within a 1⁄2 mile of the household.

The Land Use Effect of Transit: Findings 23 • Clustering future job growth in other parts of the region near high-quality transit nodes. A longer term option, improving the region-wide proximity of jobs to high-quality transit, makes living near transit a more desirable option throughout the transit network. The key findings described here can be used to predict average land use effects and land use benefits in response to transit system enhancements. Since predictions provided are averages, they will be more accurate when applied to larger improvement programs and multiple stations. It is important to keep in mind that land use effects, particularly at the local level, will vary sub- stantially in response to a number of factors, discussed further in Section 4.6. 4.5 Portland’s Westside Light-Rail Extension The datasets used in the neighborhood model provided an opportunity to conduct a parallel longitudinal analysis of actual changes in land use patterns along Portland’s Westside LRT line (western portion of the Blue Line) between 1994 and 2011. The 15-mile section, with 17 sta- tions, opened in 1998. Much of the alignment is through land that was ripe for development or redevelopment. Station areas have had many years to densify and thereby affect travel behavior. Land use changes in the light-rail corridor were compared to land use changes in a control corridor, using a statistical model. With the comparison highway corridor as a baseline, Port- land’s Westside LRT extension is associated with an increase in activity densities within the 2.5-mile catchment area of 24% and an increase in average daily transit trips per household of 60%. These changes correspond to a 6% household VMT reduction due to the land use effect and an additional 8% VMT reduction due to the ridership effect. For comparison, the other statistical models developed in this study would predict a den- sity increase of 6% in the area surrounding the Blue Line extension, given average responses seen across multiple urban areas and average levels of public support and land potential. The observed increase in activity densities of 24% demonstrates the high degree of variation in the land use effect of individual transit investments. The Westside LRT corridor identified for this test had both many sites ripe for redevelopment and one of the highest levels of government sup- port for TOD of any city in the country. The result of these factors was an increase in densities four times that of the average seen in U.S. cities. Additional detail is provided in the appendices to this report. Map and Information: City of Charlotte. SCIP = South Corridor Infrastructure Program. Figure 3. Development response patterns along the Blue Line in Charlotte, North Carolina.

24 Quantifying Transit’s Impact on GHG Emissions and Energy Use—The Land Use Component 4.6 Factors that May Influence the Land Use Effect More than just transit investments influence land use development patterns, even in areas immediately adjacent to transit. Public support and market forces play an important role in determining land use patterns. Time is another factor; new development around transit stations can happen relatively quickly, within 5 to 10 years of investments, or can happen decades later. The pedestrian environment in station areas also determines the propensity of new residents to walk and bike when they are not riding transit. There is some disagreement in the literature over how strongly and how consistently transit investments attract new development. A significant number of studies (Cervero et al. 1995, King 2011, Kolko et al. 2011) have found that transit alone does not spur new development and that other built environment features are equally, if not more, important in influencing development growth patterns. The models developed in this research use typical existing interactions between transit investments and land use patterns to predict the effects of future investments, but the results must be interpreted in the context of other factors as well. Not all of the factors discussed here can be considered in the models developed. The models predict aggregate results, at the transit system level or for groups of stations, with greater accuracy than they predict results for individual transit stations. Therefore, planners should carefully consider the potential for other factors, discussed in more detail below, to influence the land use effect, particularly where smaller geographies are of interest. 4.6.1 Public Support and Land Potential Public support for making necessary land use changes and market potential for development are the primary determinants of development in individual station areas and transit corridors. These factors impact the land use effect by influencing development densities around transit, which in turn influence the travel patterns of non-transit riders and transit riders alike. A recent study from the Institute for Transportation and Development Policy (ITDP) reviewed 21 LRT, bus rapid transit (BRT), and streetcar corridors in 13 cities across the United States and Canada to assess the effect of transit investments on development adjacent to the tran- sit corridors. Investment levels were measured in terms of dollars spent. Each corridor was rated on transit level of service (relative to the ITDP’s BRT Standard), land potential (a measure of the pre-existing attributes of a city or corridor that support development), and public support. Factors were assessed individually for their effects on land use development, using a mixture of quantitative and qualitative information (ITDP 2013). For land potential, the ITDP study found that regional market strength, as rated by Price- waterhouseCoopers, was a poor predictor of investment around transit lines. The strength of the local land market around the transit line was much more influential. ITDP classified each transit corridor’s local land market strength based on ownership, adjacent uses, topography, and availability for redevelopment. Where governments provided at least moderate support for development around transit lines, the strength of the local land market was found to be a good predictor of development levels. ITDP found a nearly direct correlation between the level of investment and the strength of government support. ITDP classified each transit corridor’s level of public support based on the level of activity in rezoning, investing in related infrastructure, land use planning, outreach to developers, providing financial incentives, environmental clean-up, land assembly, and market- ing activities. The level of transit service along transit corridors, as analyzed in the ITDP study, was the least influential indicator of development, although not inconsequential.

The Land Use Effect of Transit: Findings 25 The findings of the ITDP study are consistent with other studies in the field that have used more rigorous statistical methods. An extensive analysis of the San Francisco area’s BART heavy- rail transit system and its effects on development patterns (Cervero et al. 1995) found that the availability of vacant and developable land was an important predictor of whether land use changes occurred near stations. Local real estate markets and public support, in the form of financial incentives and assistance in land assemblage from local redevelopment authorities, played a key role in development outcomes in the first 20 years after BART’s opening. Given the importance of public support and land potential in determining the land use effect of transit, particularly in the short term, this research considered ways to quantify the effect of these factors. The research team gathered information from the CTOD National TOD Database about job and population growth in transit station areas from 2000 to 2010. The team examined growth trends with respect to the ratings developed by ITDP for various new transit corridors in terms of land potential (limited, emerging, strong) and government TOD support (weak, moderate, strong). There were no evident correlations between the ITDP ratings and observed growth patterns. There are two possible explanations for this. First, the TOD Database contains data for a limited time period, which is likely not long enough to capture the land use effects of new investments. Second, every region is subject to varying short- and long-term demographic and economic factors that affect local growth patterns independently from transit investment. The question of how to assess public support and land potential as factors in the land use effect should be the subject of future research. When applying the results of this research, planners should be aware that land use intensifica- tion around individual transit corridors, stations, and stops (and by extension, land use benefits in terms of VMT, gasoline consumption, and GHG emissions) could be higher or lower than pre- dicted by the models, due to the presence or absence of public support and market factors. For example, a separate analysis of the Westside light-rail line in Portland found a 24% increase in local densities attributable to the transit investment over a 17-year period, with a correspond- ing 6% decrease in household VMT (the land use benefit). (See Section 4.5.) This change is far higher than that predicted by the models and can be attributed to the Portland region’s strong integrated transportation and land use planning framework and a strong local market for devel- opment along the route, which combined to support relatively high building rates. Conversely, transit investments that are located in less supportive political and market environments can see zero development activity for many years. 4.6.2 Type and Quality of Transit Service The models constructed for this research suggest that the type and quality of transit service have important impacts on the land use effect of transit, even if they are not the primary factors determining development patterns. These impact the land use effect by influencing development densities around transit, which in turn influence the travel patterns of non-transit riders and transit riders alike. In one model, the average frequency of transit across the entire system has the same value in predicting land use as the density of transit service provided (in route miles per square mile). In another model, the number of jobs accessible by transit within 30 minutes has a direct effect on land use density in the local area. It follows that improving transit levels of service, and thereby increasing the number of jobs accessible within 30 minutes, would tend to increase land use densi- ties. These results suggest that level of service is just as important as having transit service available, and that increasing levels of service on existing routes may have benefits over route expansion. Traditionally, rail transit has been associated with a higher level of service, including greater reliability, separated guideways, higher speed, and shorter headways, than bus service. If typical bus headways are 20 to 30 minutes and typical rail headways are 10 to 15 minutes, one would

26 Quantifying Transit’s Impact on GHG Emissions and Energy Use—The Land Use Component expect twice the land use effect from a rail transit system as from a bus transit system. And in fact, the land use effect of existing systems with rail is nearly twice that of existing systems without rail, as shown in Table 4. While urban areas with rail service also have bus service making up a substantial share of their total transit systems, bus headways are likely to be more frequent in urban areas large enough to sustain rail service. There is some evidence that transit service, and particularly commuter rail service, can con- tribute to accelerating sprawl at the urban edge (Chatman and Noland 2013, Landis and Cervero 1999). The model used to assess land use effects of existing systems supports the notion that different types of transit service have different land use effects. More LRT is associated with higher gross population densities. The same is not true of heavy-rail transit, possibly due to the potential of rail extensions into the suburbs to promote sprawl. However, setting aside variations in land use patterns within a region, the models show that the total effect of more transit service is an increase in gross population density and a corresponding decrease in VMT. The advantage of rail over bus in generating higher land use effects may erode with the advent of BRT systems that match or even exceed the level of service provided by rail in some cases. In fact, the model results suggest that a bus system providing the same level of service as rail can generate the same land use effects. Recent studies of property development around new BRT lines have also demonstrated this potential (ITDP 2013, Nelson et al. 2011, Cervero and Kang 2011). Some have suggested that fixed-guideway transit has the potential to generate greater land use effects than non-fixed-guideway transit because the fixed infrastructure investment implies a long-term commitment by public agencies to provide transit service. The research conducted under TCRP Project H-46 finds that transit that provides higher frequency service and greater access to jobs—two qualities generally associated with fixed-guideway transit—generate higher land use benefits. 4.6.3 Vehicle Capacity While there is an obvious correlation between land use densities and the capacity of transit vehicles serving the area, providing higher capacity transit vehicles is not likely to generate addi- tional land use effects in and of itself. Figure 4 shows how different transit modes are associated with different types of development. Transit vehicle capacities tend to be higher in higher density areas. From the perspective of transit service planning, it makes sense to provide more transit capacity where more riders live and work. The statistical models in this study have illuminated three primary transit characteristics that shape the land use effect: • Transit access (represented by route density at the regional level or station proximity at the neighborhood level). • Transit frequency. • Transit employment accessibility (which captures transit speed, frequency, and network connectivity). Table 4. Average land use benefits by transit system type among sample urbanized areas. % VMT Reduction Urban Areas with Rail Service 14% Urban Areas without Rail Service 8%

The Land Use Effect of Transit: Findings 27 Source: Nelson/Nygaard. emp. = employees Figure 4. Typical land use densities associated with different types of transit. Transit vehicle capacity was not incorporated in the statistical models in this research because available data on transit vehicle capacity were not sufficiently detailed; however, it is unlikely that including a transit vehicle capacity variable would have substantively changed the model results. Literature on property value impacts of transit investments has not discussed transit vehicle capacity as a driving factor. (Property values are a reasonable proxy for land use effects because rising real property values indicate that more people want to locate in a given area, which in turn makes developing at higher densities more viable.) The economic theory behind these studies is that the improved access to destinations offered by transit drives increased property values. Transit access, speed, frequency, and network connectivity—not transit vehicle capacity—are the variables that determine access to destinations. Only consistent and severe overcrowding on transit vehicles would impact access to destinations. Transit vehicle capacity should meet the needs of the riding population in any given area. Living or working in a neighborhood may become less desirable if the transit service provided is overcrowded. But if developers believe that transit agencies will provide sufficient vehicle capac- ity to serve new development as it becomes occupied, then transit vehicle capacity should not be a driving factor in the land use effect. In other words, transit vehicle capacity should be seen as a planning decision that responds to the land use effect, rather than shapes it. 4.6.4 Road Supply Generally speaking, transit competes with the private automobile as a mode of personal trans- portation. This competition extends to the land use effect as well, where there is a discernible tradeoff between transit supply and road supply.

28 Quantifying Transit’s Impact on GHG Emissions and Energy Use—The Land Use Component Applying the models developed for this research, the research team estimates that a 1% increase in freeway lane miles per capita in an urban area is associated with a 0.1% decrease in population density. A 1% increase in non-freeway lane miles per capita is associated with a 0.5% decrease in population density in the region. 4.6.5 Time for Development Common sense suggests that time is an important factor in determining the scale of the land use effect of transit. Development happens on the time scale of decades, with multiple years needed to acquire parcels, design and finance development, acquire permits, and complete con- struction. The land use effect of transit is realized when new development occurs, bringing more residents and jobs into compact, mixed-use areas where destinations are closer together and more accessible by foot and bicycle. If new development takes decades to happen around new transit investments, the land use effect of transit will likewise take decades to be realized. The San Francisco Bay Area’s BART system is an example of this phenomenon. While some station areas attracted development in the first few decades after the transit system opened, many more station areas are seeing development only now, 40 years after the transit service opened. Much older transit stations also continue to attract development. For example, Evanston, Illinois, saw a boom in development around transit stations in the 1990s, 70 years after the transit service in question was in place (CTOD 2011). On the other hand, some cities see development that coincides with the opening of new transit or even precedes the opening of new transit lines. Phoenix, Charlotte, and Minneapolis have all seen construction projects start around their new transit lines before the lines themselves were even completed (CTOD 2011). Developers anticipated the market opportunities provided by transit access and acted early. The 2013 ITDP report cited above considered the impact of timing on the land use effects of new transit corridors. The transit corridors considered by the study have all opened in the last 10 to 20 years. ITDP found little correlation of transit system age with the amount of development adjacent to the corridors. Land potential and government support far outweighed time since opening as predictors of development (ITDP 2013). Statistical modeling conducted for this research included a longitudinal analysis of urbanized areas between 2000 and 2010 and found no land use effects during the period, suggesting that land use effects take longer than 10 years to develop after a transit investment. In Portland, an examination of development around the Westside Blue Line extension showed land use effects far higher than the effects predicted by the statistical models in only 17 years. See the following section for further details. Based on the evidence above, the research team concluded that time has a highly unpredict- able relationship to the land use effect. It is reasonable to expect a minimum of 10 years for land use development around transit to occur, but it may take many more years. The importance of government support and market factors in determining the rate of development cannot be understated. To make more accurate predictions of timeframes for development in individual regions, planners should consult historical development data for their region or conduct a mar- ket forecast study for the neighborhood or corridor of interest.