Recent Decline in Public Transportation Ridership: Analysis, Causes, and Responses (2022)

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.

78 BRT, or bus rapid transit, is defined by FTA as “a high-quality bus-based transit system that delivers fast and efficient service that may include dedicated lanes, busways, traffic signal priority, off-board fare collection, elevated platforms, and enhanced stations” (Federal Transit Adminis- tration, 2015). This definition suggests that BRT is a combination of different features. Transit agencies follow different approaches when planning and implementing BRT routes (Levinson et al., 2003a). Some transit agencies implement nearly all BRT features available, while others choose to implement only some of the features. Converting local bus routes to BRT is a strategy used by transit agencies to provide riders with high-quality service without the intensive capital investment required for light and heavy rail. Over the past several decades, the Greater Cleveland Regional Transit Authority (GCRTA) con- verted three of its highest ridership bus routes into BRT lines, with different BRT features that range from all BRT features to just branding, as detailed in Section 9.2. This section describes research to quantify the ridership impacts of converting local bus routes to BRT. Assessing the impacts of BRT on ridership will inform transit agencies and city officials who are planning new BRT routes or converting existing bus routes to BRT. 9.1 Objective of BRT Analysis The research evaluated bus ridership in Cleveland, Ohio, for the period of 2004 through 2019 in an effort to assess the following research questions: • How has converting a local bus route to full BRT impacted ridership? • How has converting a local route to have some BRT features impacted ridership? • How has branding a local route as BRT without other enhancements impacted ridership? 9.2 BRT Routes in Cleveland The transit agency in Cleveland operates three BRT routes: the HealthLine, the Cleveland State Line, and the MetroHealth Line. These were launched in 2008, 2014, and 2017, respectively, as shown in Figure 9-1. Each of these three routes has a unique set of BRT features; this presents a unique opportunity to investigate the impact of different levels of BRT features on bus ridership. More details about the features of each BRT route are discussed in the following subsections. 9.2.1 HealthLine BRT Route In October 2008, GCRTA opened the HealthLine, a state-of-the-art BRT line replacing Bus Route 6 along Euclid Avenue. The HealthLine was designed in part to support the Euclid Corridor C H A P T E R 9 The Impact of Converting Bus Routes to BRT on Ridership in Cleveland, Ohio BRT is “a high- quality bus-based transit system that delivers fast and efficient service that may include dedicated lanes, busways, traffic signal priority, off-board fare collection, elevated platforms, and enhanced stations.”

The Impact of Converting Bus Routes to BRT on Ridership in Cleveland, Ohio 79   Transportation Project, a multiyear project implemented to revitalize Euclid Avenue as a business district through transit-oriented development along the corridor. According to GCRTA (2020), the HealthLine has the following BRT features: • 4.5 miles of dedicated right-of-way • High-frequency service (every 10 minutes during peak hours, 15 minutes during off-peak) • 24/7 bus service • TSP • Real-time information displays • Bus stop consolidation • Level boarding stations • Off-board fare collection • Unique branding This conversion of the HealthLine BRT route reduced the average running time from 46 minutes to 36 minutes (Federal Transit Administration, 2012). However, it should be noted that shortly after opening the HealthLine, TSP for portions of the HealthLine’s route was discontinued. Additionally, there were policy issues related to the off-board fare collection. GCRTA’s fare evasion policing was deemed unconstitutional by a Cleveland Municipal Court judge in October 2017, which resulted in GCRTA switching to onboard validation and requiring riders to pay for or validate a ticket on the bus itself (Allard, 2019). It also worth noting that the HealthLine is the only route that FTA has accepted for reporting as the BRT mode in the NTD. 9.2.2 Cleveland State Line BRT Route In December 2014, the Cleveland State Line BRT route opened, connecting downtown Cleveland—including Cleveland State University—to the western part of Cuyahoga County. The Cleveland State Line enhanced the 55 family of bus routes. The Clifton Boulevard portion of the Cleveland State Line has the following BRT features: • Dedicated transit lane for both buses and bikes during peak hours (7:00–9:30 a.m.; 4:00–6:30 p.m.) • Improved traffic signal system for all vehicles • New stations • Bus stop consolidation • Median landscaping • Enhanced streetscape • Unique branding 9.2.3 MetroHealth Line BRT Route In December 2017, GCRTA rebranded its second-busiest route as the MetroHealth Line BRT route. The MetroHealth Line BRT is a rebranding of the 51 family of bus routes. This BRT route Cleveland State Line BRT Route (Some BRT) Oct 2008 Dec 2014 Dec 2017 HealthLine BRT Route (Full BRT) MetroHealth Line BRT Route (BRT Branding) Figure 9-1. The City of Cleveland BRT routes launch dates.

80 Recent Decline in Public Transportation Ridership: Analysis, Causes, and Responses opened with limited changes, namely rebranded buses (see Figure 9-2), refurbished shelters, and customized signage (MetroHealth, 2017). There are plans to implement TSP and dedicated lanes during peak hours on the MetroHealth Line, but these features had not yet been implemented at the time of writing. 9.3 Data Sources This study used data from different sources, as shown in Table 9-1. The unlinked bus trips, VRH, and BRT features were obtained from GCRTA. Other variables—like employment, gas prices, percent of zero-vehicle households, and years since the introduction of TNCs—were obtained from publicly available data sources. 9.4 Ridership Trends This section discusses ridership trends for BRT routes and non-BRT routes in Cleveland. Fig- ure 9-3 shows that after the launch of the HealthLine (which has full BRT features) in 2008, the route continued to gain ridership until it peaked in 2014. After 2014, ridership on the HealthLine started to drop. The Cleveland State Line (with some BRT features) also gained ridership after its launch in 2014 and has not suffered a significant decline in ridership. However, the MetroHealth Line (with only BRT branding) did not gain ridership after its launch in 2017 but continued a downward ridership trend similar to the non-BRT routes (see Figure 9-3). Furthermore, the ridership of the three BRT routes declined in 2018 and 2019. Source: http://www.riderta.com/news/dec-7-2017-metrohealth-line-buses-debut-week Figure 9-2. MetroHealth Line bus. Variable Data Source Unlinked bus trips GCRTAVRH BRT features Employment Bureau of Labor Statistics Gas prices Energy Information Administration Percent of zero-vehicle households One-year American Community Survey Years since the introduction of TNCs Uber and APTA Table 9-1. Data sources for Cleveland BRT analysis.

The Impact of Converting Bus Routes to BRT on Ridership in Cleveland, Ohio 81   The declines in BRT ridership appear to be similar to general ridership declines occurring in Cleveland since 2014. These declines could be attributed to some internal factors, like reducing service provision and increasing fares, as well as other external factors that affect ridership, such as changing gas prices and the introduction of TNCs. More details about changes in these factors are discussed in the following subsections. 9.4.1 Internal Factors Affecting Ridership The first contributing factor to ridership declines appears to be a drop in the amount of service provided. Figure 9-4 shows that VRH declined for all three BRT routes in 2018 and 2019, and service reductions for non-BRT routes started in 2016. This is expected to reduce ridership since service provision has been identified as an important determinant of transit ridership based on prior studies (Boisjoly et al., 2018; Evans et al., 2004). The second potential contributing internal factor is fare increases. Figure 9-5 shows that the average fare per UPT increased from 2016, which is when GCRTA implemented a 25-cent fare increase (GCRTA, 2012). Many prior studies have shown that fare increases often result in rider- ship declines (Taylor et al., 2009; McCollom and Pratt, 2004). H ea lth L in e L au nc h C le ve la nd S ta te L in e L au nc h M et ro H ea lth L in e L au nc h Figure 9-3. Annual bus ridership for BRT and non-BRT routes in Cleveland. C le ve la nd S ta te L au nc h H ea lth L in e L au nc h M et ro H ea lth L in e L au nc h Figure 9-4. Annual bus service provision for BRT and non-BRT routes in Cleveland.

82 Recent Decline in Public Transportation Ridership: Analysis, Causes, and Responses 9.4.2 External Factors Affecting Ridership In addition to the internal factors, there were changes in some external factors that were expected to reduce bus ridership in Cleveland. Figure 9-6 shows the percent of zero-vehicle households, gas prices, and percent of workers who telecommuted in Cleveland for the period of 2004 to 2019. First, it can be noticed that the percent of zero-vehicle households has decreased since 2014; this is likely to have a negative impact on transit ridership since more residents have access to automobiles. Second, gas prices aer adjusting for ination are cheaper now compared to 2014; this is also expected to negatively impact transit ridership. Furthermore, telecommuting has increased in Cleveland since 2014, which may also have negative eects on transit ridership. Finally, the introduction of TNCs in 2014 could have reduced transit ridership (Graehler et al., Source: NTD. Figure 9-5. Fare per UPT. In tr od uc tio n of T N C s Figure 9-6. Gas prices, percent of zero-vehicle households, and telecommuting in Cleveland.

The Impact of Converting Bus Routes to BRT on Ridership in Cleveland, Ohio 83   2019). In summary, all these changes in external factors were expected to cause ridership drops and, as shown in a previous chapter, have impacted ridership in our own analysis. 9.5 Results of the BRT Analysis Using the fixed-effects regression techniques described in Appendix G of TCRP Web-Only Document 74, this study assessed the impacts of BRT routes on ridership over the period from 2004 to 2019. The dependent variable was the total bus trips per route per year. It should be noted that the log trans formation of the dependent variable was used; therefore, some of the results of the models can be interpreted as ridership elasticity values. The explanatory variables were the differ- ent BRT levels, as described in Section 9.3, and other commonly used explanatory variables such as transit VRH, employment, percent of zero-vehicle households, and years since the introduction of TNCs. The results of two preferred models with similar statistical properties are in Table 9-2. The results of this analysis suggest that the elasticity of ridership with respect to changes in transit VRH is about 0.9, as shown by both Model 1 and Model 2 in the first row of Table 9-2. This value suggests that increasing/decreasing VRH on a route by 10% will increase/decrease route ridership by 9%. This finding is consistent with the nationwide analysis conducted in this study that found the elasticity of ridership with respect to changes in transit service provision to be about 0.83. Table 9-2 also shows that converting bus routes to full BRT or adding some BRT features resulted in ridership gains that ranged between 22% (based on Model 2) to 46% (based on Model 1), as shown in the second and third rows of Table 9-2. It should be noted that the coefficient values for full BRT and some BRT features were consistently positive and significant in the model results, which provides evidence that these two routes increased ridership (holding all else equal). However, the value of these coefficients varied some depending on the model specification. Given that a simple dummy variable (i.e., binary variable) was used to represent BRT, the research team sug- gests that additional research be conducted in the future to help untangle how specific features of BRT impact ridership. Last, just branding routes as “BRT” did not have a significant impact on ridership, as shown by the MetroHealth Line results in Model 1 and Model 2 in Table 9-2. The results of this study also suggest that total employment and the percent of zero-vehicle households have significant positive impacts on bus ridership, while TNCs have significant Dependent Variable: Unlinked Bus Trips per Route per Year (Log Transformed) Model 1 Model 2 VRH (log transformed) 0.88*** 0.91*** HealthLine (full BRT features) 0.31*** 0.22*** Cleveland State Line (some BRT features) 0.46*** 0.42*** MetroHealth Line (only BRT branding) 0.03 0.002 Percent of zero-vehicle households - Significant positive effect Years since the introduction of TNCs Significant negative effect - Total employment Significant positive effect Route Controlled for differences between routes Number of observations 697 Variable significance: ***p < 0.01; **p < 0.05; *p < 0.10; no star = not significant Table 9-2. Impact of BRT on unlinked bus trips model results.

84 Recent Decline in Public Transportation Ridership: Analysis, Causes, and Responses negative impacts on bus ridership. These findings are consistent with the nationwide analysis conducted in this study as well as prior studies. It should be noted that TNCs and the percent of zero-vehicle households were evaluated separately (in Model 1 and Model 2, respectively) due to multicollinearity; for this reason, two separate models with similar results are shown in Table 9-2. 9.6 Conclusions, Discussion, and Implications of the BRT Analysis This study conducted an empirical analysis to quantify the impacts of converting local bus routes to BRT using Cleveland as a case study. The results of this study suggest that converting bus routes to full BRT or adding substantial BRT features increased ridership significantly (possibly between 22% and 46%) per year. These results suggest that giving buses priority on streets yielded ridership gains even if this priority was limited to peak hours. However, merely branding routes “BRT” did not result in significant changes in ridership; this is likely because branding does not improve the level of the service. These findings are important for transit agencies that are considering increasing the priority of bus routes on local streets. It is also important to note that the BRT routes in Cleveland experienced net ridership declines in 2018 and 2019. These decreases appear to be part of general transit ridership declines in Cleveland that have occurred over the past five years (since 2014) and are similar in nature to overall down- ward trends of transit ridership in the United States. In Cleveland, these decreases are likely driven by both internal and external factors discussed in a previous section. This includes decreases in the amount of transit service provided, fare increases, the introduction of TNCs, and reductions in the cost of owning and operating an automobile. In summary, the positive ridership impacts attributed to BRT in the previous paragraph were not sufficient to outweigh the negative effects of other factors that have impacted ridership in Cleveland, resulting in net ridership declines. Converting routes to BRT resulted in bus ridership increases of 22% to 46%.