The Exposition Light Rail Line Study

A Before-and-After Study of the Impact of New Light Rail Transit Service

Principal Investigators
Marlon G. Boarnet
Doug Houston (UC Irvine)
Research Team
Andy Hong, Jeongwoo Lee, Xize Wang, Weijie Wang, and Steven Spears (UC Irvine)

Download the full report (PDF) || Appendix section (PDF) || Read the USC News media release

Executive Summary

Background
Los Angeles, the world’s prototypical automobile city, is transforming into a multi-modal metropolis. The six rail transit lines projected to open between 2012 and 2020 will make the Los Angeles Metro Rail system longer than the present day Metro in Washington D.C. At the same time, ambitious state regulations require that metropolitan planning organizations demonstrate how their transportation plans meet greenhouse gas emission (GHG) reduction targets. There is a pressing need to evaluate the impact of new transportation investments comprehensively. Yet, transportation, as a field, has rarely systematically evaluated the impact of major projects using an experimental – control group design. To help close that gap, we conducted the first-ever experimental – control group, before – after study of the impact of a major transportation investment in California.

As transportation becomes more varied – with localities experimenting with programs that include real-time parking pricing, toll lanes, neighborhood vehicles, and bicycle plans – it will be increasingly important to evaluate the impact of these projects in a consistent and credible way. Social scientists have applied the methods of experimental research designs for decades, but such techniques have only recently made inroads in transportation.

Study Design
We conducted a detailed study of travel behavior changes around new Expo Line light rail stations, using experimental methods. The Exposition (Expo) Line is a light rail line in the Los Angeles metropolitan area that extends south and west from downtown Los Angeles. Phase I of the line, which opened in two stages in April and June 2012, runs 8.7 miles from downtown Los Angeles westward to Culver City, near the junction of the 405 and 10 Freeways. The six western-most stations along the Expo Line (Phase I) comprise the experimental neighborhoods, and similar control group neighborhoods were chosen nearby. This research project enrolled experimental households, within ½ mile of a new Expo Line station, and control households, living beyond ½ mile from the station. In fall of 2011, those households were asked to track their travel for seven days, recording daily odometer readings for all household vehicles and logging trips by travel mode and day for each household member 12 years or older. In approximately half of the households, an adult also carried a geographic positioning device (GPS) and an accelerometer, to measure travel (via the GPS device’s location tracking function) and physical activity. The same households were invited to complete the seven day travel study again in fall, 2012, after the Expo Line opened. In total, 204 households (103 in the experimental neighborhoods, 101 in control neighborhoods) completed the travel tracking before and after the Expo Line opened.

We used those data to conduct before-after evaluations of the impact of the Expo Line on travel behavior. The research design is a classic “differences-in-differences” approach. The impact of the Expo Line can be inferred by examining experimental minus control group differences and how those differences change after the Expo Line opens relative to baseline, “before opening” experimental minus control group differences. The travel behavior variables studied are household daily averages for: vehicle miles traveled (VMT), car driver trips, train transit trips, bus transit trips, walking trips, bicycle trips, walking minutes, and bicycling minutes. Additionally, we studied changes in physical activity for the adults who carried an accelerometer, and CO2 emissions for household vehicles.

Results

The analysis gives the following results.

• In “before opening” travel data collection, experimental and control households had the same travel patterns. There were no statistically significant differences across experimental and control households, before the Expo Line opened, in household daily average VMT, car driver trips, train transit trips, bus transit trips, walking trips, bicycle trips, walking minutes, or bicycling minutes.

• After opening, the differences-in-differences approach shows that the experimental group reduced their daily household VMT by 10 to 12 miles relative to the control group. That result persists after outlier observations are removed and when alternative statistical methods are used. We interpret this as evidence that the Expo Line reduces VMT among households living within ½ mile of the Expo Line stations.

• We used the GPS data to examine whether the large VMT changes could be an artifact of households systematically misreporting vehicle odometer logs. While the GPS provides at best a crude check on odometer logs (because the person carrying the GPS could have ridden in several vehicles), we find no evidence of any systematic reporting biases that would reduce our faith in the result that experimental households reduced their VMT by 10 to 12 miles, relative to control group households, after the Expo Line opened.

• In some statistical tests, there is evidence that the Expo Line increased rail transit ridership among experimental households. Control group households also increased their rail ridership, but not by as much as experimental households. On net, the differences-in-differences evidence suggests that the Expo Line resulted in about 0.1 more daily train trips per household in the experimental group, but we caution that this result is not nearly as robust as the finding for VMT reduction among experimental group households.

• The experimental and control group households had no statistically significant differences in vehicle CO2 emissions before the Expo Line opened, but after opening experimental group households had approximately 30% less vehicle CO2 emissions than control group households. That “after opening” difference is statistically significant.

• The accelerometer data allow us to measure physical activity in minutes of moderate or vigorous activity per day. After the Expo Line opened, those individuals living in the experimental neighborhoods who were the least physically active had the largest increases in physical activity relative to control group subjects. The Expo Line opening was associated with increases in physical activity among approximately the 40 percent of experimental subjects who had the lowest physical activity levels before the line opened. The impact was as high as 8 to 10 minutes of increased daily moderate or vigorous physical activity among those experimental group subjects who were the least active before the Expo Line opened. Note though that for more than half of the experimental group subjects (those more physically active before the Expo Line opened) our statistical test suggests that the Expo Line is associated with decreases in physical activity.

• The impact of the Expo Line on VMT and rail ridership was larger near stations with more bus lines and near stations with streets with fewer traffic lanes, suggesting that bus service increases the impact of rail transit and that wide streets (which can be barriers to pedestrian access) reduce the impact of rail transit, at least in the Expo Line corridor.

Summary and Policy Implications
Los Angeles has made a substantial commitment to rail transit, but several policy questions continue to be debated. Among those questions, possibly the most basic is whether new transportation options will change travel modes, and whether Angelenos will modify their travel as new options become available. Against that backdrop, the results from this research are in some ways striking. We find evidence that the Expo Line is associated with large reductions in VMT, some increase in rail transit ridership, changes in physical activity, and large reductions in GHG emissions among households living within ½ mile of a station. The research design, using a control group to account for factors other than the rail investment, allows us to make causal inferences more strongly than is often the case in social scientific research. In short, the Expo Line is associated with travel behavior change, and we can infer that the association reflects a causal effect of the Expo Line on household travel. The large impacts for VMT and GHG reduction occur within a small area – ½ mile around six new stations. Viewed from the perspective of the greater Los Angeles region, these impacts will be small, but they are large in the neighborhoods surrounding the Expo Line.

Several policy implications follow. First, this is some optimistic evidence for the rail transit investment program in Los Angeles. We did not conduct a formal benefit-cost assessment, but clearly the rail line is associated with changes in travel behavior that are consistent with the anticipated effects. Second, the evidence indicates that the travel impacts of light rail are enhanced by local land use characteristics. Bus lines and streets with narrower width are likely to be more conducive to increasing the effects of rail transit, at least based on the results from this study. Third, economic theory predicts that the travel impacts documented here will likely lead to downstream effects including changes in the resident population near Expo Line stations, changes in land prices (and hence rents and house prices), and changes in land uses. It is too early to draw conclusions about those downstream effects, and whether they will advantage existing residents or new residents or a combination of both. Yet, the fact that the Expo Line has changed travel behavior suggests that the rail service is valued by nearby residents. That value will likely be reflected in land prices and land uses in later years, which will lead to broader social and economic impact around the rail transit lines.

Download the report (PDF) || Appendix section (PDF)

Research Funding
Funding for the study was provided by the Haynes Foundation, the Lincoln Institute of Land Policy, the Southern California Association of Governments, the University of California¹s Transportation Center and Multi-Campus Research Program on Sustainable Transportation, and the USC Lusk Center for Real Estate.