Evaluating Light Rail Rebirth in the United States
Light rail transit has become popular among policy makers and planners as the public transit option to address growing concerns of congestion and to spur economic development. These benefits of light rail are contingent on high transit utilization. By evaluating a sample of 16 cities with modern light rail systems, this research finds that the success of a system is largely dependent on the service area population and the extent of the system. However, this research also shows that large system expansions do not always result in high return in terms of increased riders. Thus, light rail may not always be the best option for every urban center, and it may not yield the overreaching goals light rail supporters claim.
Light rail transit has existed in some fashion for over 100 years, but it was not until 1989 that the Transportation Research Board adopted a formal definition. They define light rail transit (LRT) has “a metropolitan electric railway system characterized by its ability to operate single cars or short trains along exclusive rights-of-way…in subways, or occasionally, in streets…” (Transportation Research Board, 2001). In essence, LRT differentiates itself from traditional subways by having shorter trains and sometimes utilizes a mixed guide-way to merge with car traffic. There has been a rebirth of LRT in cities across the United States since the 1980s, with one new system on average opening service each year. Supporters of LRT champion these systems as a way to alleviate congestion, generate economic development, and improve mobility. Congestion relief and economic development are two outcomes that are dependent on people using LRT. There can be no congestion relief if people do not switch from driving their car to riding the train. Similarly, economic development relies on the LRT improving accessibility to economic centers, such as business districts and transit oriented developments, which is normally reflected in growing ridership.
Since most benefits are outcomes of LRT utilization, this paper sets out to analyze ridership trends and performance and service measures. Strong performing LRT systems attract a higher ridership for the amount of service they provide when compared to their peers. By assessing which cities are successful versus the ones which are not, this research seeks to find the differences between the two groups. The goal is to find the most important system and city characteristics that determine successful LRT systems.
This research can be broken down into four overarching questions:
- How are LRT systems performing against each other?
- How does LRT compare with other modes of travel?
- How have LRT systems changed as they were built?
- What are the biggest factors driving ridership?
This research focuses on modern light rail, which I define as systems built after 1980. From there, cities were selected to have a distribution of LRT systems from across the US. Only systems with significant ridership were considered, so LRT systems with less than approximately 1 million annual unlinked passenger trips were excluded. The table below contains the 16 cities used in this research. Provided in the table is the name of the governing transit authority and the first year of service.
How Are LRT Systems Performing Against Each Other?
Answering this question requires a definition of system performance. There are many metrics used by transit experts to rate system performance. The key metrics used are as follows:
- Ridership: Measured in annual unlinked passenger trips (via vehicle boardings)
- Directional Route Miles (DRM): Geographical extent of system
- Revenue Vehicles Miles (RVM): Density of service
- Service Area: The area enclosed by the borders of the governing transit authority
- Service Area Population: The population residing within the service area
- RVM/DRM: Represents intensity of service provided
- DRM/Service Area: Represents system coverage
The first metric to consider is ridership trends. Ridership patterns over time illustrate the differences between growing, stable, and declining systems. The ridership values were normalized by setting the initial value equal to 100 to make fair comparisons among systems of all sizes. Using American Public Transportation Association (APTA) data, the graph below shows the ridership for each LRT starting from year zero, which represents their respective opening year of service.
In Figure 1, it is clear some cities have had incredible ridership growth since opening. Systems that stand out in particular are Seattle, Salt Lake City, Dallas, Charlotte and Minneapolis. However, there is a scaling issue with the graph because ridership in Seattle and Salt Lake City has risen so rapidly. As a result, it is hard to determine what the other cities’ ridership changes look like. To remedy this issue, the graph below is the same graph as in Figure 1, except the vertical axis was restricted to create a better image.
Figure 2 shows more clarity on the differences between the other systems. Ridership in Buffalo appears to be declining, while Pittsburgh maintains stable ridership. The other cities show ridership growth but at varying rates.
Ridership is a major performance measure. Since transit authorities are in the business of “selling” trips, ridership levels are reflective of demand. That said, ridership is one of many metrics that must be considered. The graph below illustrates various other service metrics from the National Transit Database (NTD). In order to plot metrics with different units on the same graph, all of the measurements were normalized to remove units. On each measurement in consideration, the 16 systems were ranked from highest value to lowest value. For any particular metric, the system ranked number one was given a value of 1, the lowest system was given a value of 0 and the remaining systems in the middle were scaled in proportion. The results are displayed in Figure 3.
The initial hypothesis was that there would be a “clumping” of data points for each city; this means that for a given city, all five metrics considered would be grouped tightly together. The reasoning is that all these metrics are interconnected, and poor performance on one would lead to poor performance on the others and vice versa. In some places, like Baltimore, there is a clear gathering of the metrics at the low end. However, cities like Los Angeles and Houston have a wide range of their performance on each metric.
From this analysis, there are visible “front runners” for successful systems, among which are; Los Angeles, Denver, Houston, Seattle, and Minneapolis. Some of the weaker performing systems are Buffalo, Baltimore, and Pittsburgh.
How Well Does LRT Compete With Other Modes?
In the vast majority of US cities, the primary public transit mode is bus2. It is possible that when a new LRT system is built in a city, there is a mode shift from people riding buses to riding light rail. Growth in LRT ridership does not facilitate congestion relief, however, if the new riders are only coming from buses and not from cars. Therefore, it is crucial to look at how LRT ridership grows alongside bus ridership.
Figure 4 shows the ridership in Salt Lake City. This is an example of a promising system. When the light rail came online in 1999, bus ridership maintained steady levels, and it remained stable while light rail rose.
In Houston, as shown in Figure 5, there appears to be a less than ideal situation. Bus ridership has been declining since 2001. Light rail opened in 2004 and appears to bump up total transit ridership. However, bus ridership again dips down after 2009 and light rail increases slightly, allowing total transit ridership to stay stable. It is unclear how LRT has impacted bus ridership, but Houston does not display a great jump in total ridership that Salt Lake City did. The other 14 cities showed a pattern either like Salt Lake City or like Houston. No city had LRT ridership go up and bus ridership decline dramatically. This implies that there is no general pattern of LRT stealing ridership from buses, which would lead to little or no congestion relief.
Unfortunately, this analysis is only based on number of trips on transit. One would need to conduct a survey of LRT riders to ask them which mode of travel they used before the light rail opened in order to properly assess where the riders are coming from.
How Have LRT Systems Changed as They Were Developed?
In some cases, like Buffalo, the light rail system was built as a complete system from its opening. However, other cities built their system in segments. Understanding the productivity of this incremental growth is useful to know if system expansions are worth their cost. There are two contradictory hypotheses with system expansion. Steve Polzin states that some argue a system expansion opens up more access through more stations and thus would result in more ridership. Polzin also points out that the initial segment of a line is placed in the busiest transit corridor, so more expansions would be in a less desirable location for public transit (2003).
Figure 6 and Figure 7 highlight two common trends found among different properties. In Los Angeles, as the system expanded (DRM grew), the number of trips and trips per DRM (a measure of productivity of each route mile built) also grew. In Salt Lake City, a similar trend is seen at the beginning, but the expansion in 2009 saw trips per DRM drop significantly. While Salt Lake City did see ridership grow, the growth was not in proportion with the size of the expansion. This suggests that simply building more track does not guarantee a response in ridership growth.
What Are the Biggest Factors Behind Ridership?
This is perhaps the ultimate question to ask. High ridership is critical in obtaining congestion relief and economic development. Knowing the underlying drivers of ridership can give insight into LRT planning. A multivariate regression model was constructed to understand how DRM and service population impact ridership. Using a confidence level of 95%, these metrics proved to be statistically significant in the regression model.
Figure 8 shows a plot of the predicted ridership from the regression model versus the actual ridership. The blue dashed line shows the linear “best-fit.” This graph shows how well the model fits the observed data. From this, it can be said that cities above the line attract riders at a rate that is more than proportional to the service area population and extent of the system. This model does not indicate the specific causes for the higher ridership.
There are limitations to the effectiveness of this regression model as there are other variables that may be important in determining ridership that cannot be easily measured. However, it is still a useful tool to understand the important variables that yield ridership.
Not all LRT systems in the US are doing well with regards to ridership. As a result, the promised claims of congestion relief and economic development may not always come to fruition. Broad research of transit systems like this often oversimplifies how transit is used. The decision to use LRT as a traveler is made by individuals, who are shaped by cultural and social forces, which cannot be captured in simple quantitative metrics. Ultimately, people should be wary of claims that LRT will be the transit savior. While it has shown great success in some areas, it is difficult to make generalized predictions about how successful the next LRT system will be.
This research was made possible by the Northwestern Undergraduate Summer Research Grant. A special thank you to Dr. Joseph Schofer for his guidance on this project.
American Public Transportation Association. (2015). Public Transportation Ridership Report. American Public Transportation Association, Washington DC.
National Transit Database. (2014). Transit Agency Profiles. Federal Transit Administration.
Polzin, Steve. Ridership Trends of New Start Rail Projects. National Center for Transit Research. 2003.
Transportation Research Board. 2001. This is Light Rail Transit. (Number E-C033).