San Francisco Bay Area CONGESTION TRENDS REPORT

Transcription

1 San Francisco Bay Area CONGESTION TRENDS REPORT AUGUST 2018

2 The analysis in this report was conducted by Transpo Group using data provided by INRIX. The only Transportation Network Company that provided data for this report was Lyft. The conclusions of this report do not necessarily reflect the official policy positions of Lyft, INRIX, Transpo Group, or the Cities of San Francisco, San Jose or Oakland.

3 Table of Contents Executive Summary...1 Introduction...3 Methodology... 7 Travel Demand & Capacity Drivers Performance Trends...19 Findings & Conclusions... 33

4 THIS PAGE INTENTIONALLY LEFT BLANK

5 List of Figures Figure 1: San Francisco Bay Area Study Map...6 Figure 2: Study Methodology...8 Figure 3: Gross domestic product (GDP) by metropolitan area ( ) Figure 4: San Francisco Bay Counties Employment Rate ( ) Figure 5: 2017 Driver Population and Vehicle Use Figure 6: Gasoline and Diesel Prices (2014 to 2018) Figure 7: Quarterly U.S. E-commerce (2014 to 2017) Figure 8: San Francisco Mode Share Change ( ) Figure 9: San Francisco & Outside San Francisco Mode Share (2017) Figure 10: San Francisco Bicycle Ridership ( )...17 Figure 11: San Francisco Bicycle Network (2016)...17 Figure 12: Average Weekday Passenger Boardings ( ) Figure 13: Segment Speed & Travel Time Changes...22 Figure 14: Average Speed in San Francisco Study Area ( )...23 Figure 15: Average Speed in San Jose Study Area ( )...23 Figure 16: Average Speed in Oakland Study Area ( )...23 Figure 17: Lyft Trips Growth vs Speed Change in Study Areas ( ) Figure 18: Average Number of Lyft Trips by Time of Day (October 2017) Figure 19: Average Daily Lyft Pickups in October Figure 20: Average Daily Lyft Drop-Offs in October Figure 21: Average Speed and Lyft Activity on Geary Blvd Between Mason St and Leavenworth St, Fridays in October Figure 22: Average Speed and Lyft Activity on Taylor St Between Turk St and California St, Fridays in October Figure 23: Average Speed and Lyft Activity on Sutter St Between Stockton St and Larkin St, Fridays in October Figure 24: Lyft Line Trips in San Francisco Bay Area ( ) List of Tables Table 1: Congestion Trends Report Data Sources...9 Table 2: Average Number of Daily Pickups and Drop-Offs near Caltrain Stations (October 2014 & 2017)...25 Table 3: Top Bay Area Lyft Origins and Destinations Outside San Francisco (October 2017) Table 4: Friday Lyft Pickups & Drop-Offs (October 2017)... 28

6 THIS PAGE INTENTIONALLY LEFT BLANK

7 Executive Summary The evolution of Transportation Network Companies (TNCs) in the recent past has led major cities to attempt to evaluate the impacts of TNC services on mobility, equity, and congestion. The results of various studies concerning TNC contributions to congestion have been mixed. A recent evaluation of Lyft operations in New York City revealed that Lyft was not the primary contributor to the growth of congestion between 2014 and Transpo Group sought to understand how recent and emerging trends in the San Francisco Bay Area might shed new light on the growth of TNC services and their contributions to congestion management. Understanding this issue requires acknowledgment that congestion and transportation system performance is a complex multi-variable problem. Congestion is driven by numerous factors including changes in demand, capacity, and mode split. Prioritization of bus, bicycle, and pedestrian activity can also increase congestion for motor vehicles, particularly when travel and parking lanes are eliminated. In many cities, TNCs perform a vital function by reducing the demand for parking and connecting people to public transportation. KEY FINDINGS This study explored the drivers of congestion and evaluated various aspects of transportation system performance in the Bay Area. Transpo Group Bay Area locations with the largest growth in Lyft activity have not had the largest decrease in average daily speeds. Areas with little Lyft growth have seen more dramatic increases in congestion, suggesting that Lyft may not be the primary factor in congestion growth in the Bay Area. researchers identified various publicly-available data sources related to economic growth, population and demographic trends, and transportation system management. After evaluating the drivers of demand, we obtained Lyft data in order to understand how Lyft serves Bay Area trips. San Francisco Congestion Trends Report August

8 Approximately 70 percent of Lyft trips in the Bay Area are intracity trips, meaning they start and end within the same city and do not add to the growing congestion on key regional roadways. Our evaluation of the congestion drivers revealed that trip growth is being driven by economic activity, justified by the following findings: Population is increasing while the fraction of employed people in the region is also growing, meaning more people are traveling to and from work. Long-distance commuting in the Bay Area has increased as a result of a lack of residential development near employment centers. Private vehicle-based commute trip share has decreased, while non-private auto mode share (including TNCs) has increased slightly. Speeds are declining throughout the Bay Area but fell most in downtown San Jose and Oakland/Berkeley, less in downtown San Francisco and along major highways. Approximately 70 percent of Lyft trips in the Bay Area are intracity trips, meaning they start and end within the same city; they do not add to the growing congestion on regional roadways. Lyft trips account for less than 1 percent of daily traffic along the San Mateo, Golden Gate, and Bay Bridges. Lyft pickups and drop-offs outside of San Francisco are concentrated around Caltrain and BART stations, supporting other observations that TNCs are serving as a first/last mile connection to transit. Approximately 50 percent of Lyft trips in the Bay Area occur outside of San Francisco. Lyft Line now serves approximately 40 percent of all Lyft trips in the Bay Area. 2 August 2018 San Francisco Congestion Trends Report

9 Introduction The growth of transportation network company (TNC) services, such those provided by on-demand, shared-ride mobility platforms, has enabled new economic opportunities and travel options for people throughout cities all over the world. Access to reliable transportation at a reasonable, known, and agreedupon cost has aided people in a wide variety of economic situations. TNCs have improved access for disadvantaged communities, people who live in areas under-served or not served by transit, and people who seek to reduce their impact on carbon emissions by choosing to not own a car or occupy valuable space on congested urban streets by parking a car. TRANSPORTATION PERFORMANCE MODELING IS COMPLICATED The very nature of TNC operations, however, is that TNC ridership growth generally leads to more TNC vehicles operating on the roadway network. This growth must be understood in the overall context of congestion management, however, and may not be the sole contribution or even a major contribution to increased congestion. Understanding how TNC network growth affects congestion requires an independent and objective assessment of transportation system performance. Objectivity is achieved in such a study through the use of widelyavailable, public agency data, aggregated and anonymized locationbased data that is inherently objective, and analysis of changes in TNC use against a baseline condition for TNC operations. It is clear from extensive research that many factors influence the growth of congestion. Simply examining traffic volume, average speeds, and travel times can indicate the performance of a facility but such an San Francisco Congestion Trends Report August

10 This study was prepared by Transpo Group, a consulting firm with expertise in transportation planning and traffic operations. The study team included transportation modelers with extensive experience in simulation modeling and performance analysis, transportation planners, a traffic operations and safety engineer, data scientists, and senior performance analysts. The team members developed and refined the study methodology, selected and vetted data sources, performed data verification and quality management activities, and conducted the analysis, reviewing the outcomes in a multidisciplinary team focused on holistic transportation system performance. 4 examination cannot address complex socio-economic factors that lead to increased demand. Any effort to thoroughly understand the drivers that influence travel demand must look beyond pavement, streets, and the transportation network. Even as the scope of a study expands to captures these drivers of demand, a comprehensive examination of transportation management policies must also be undertaken to assess the impacts of speed management, vulnerable user accommodations, curb use and parking, and the management of events, closures, and traffic signal operations. PREVIOUS TNC/CONGESTION RESEARCH Many previous research efforts have attempted to quantify the TNC contribution to growing congestion in cities across the country. These reports have used differing methodologies; some using VMT as a method to quantify congestion, some using surveys to evaluate what mode people are leaving in favor of TNCs, and others using demographic data to understand who is using TNCs. The results of these studies often conflict, some determining congestion is rising because TNC VMT is increasing (Shaller, 2017), another concluded that TNCs have the potential to reduce overall traffic congestion (Li, et al, 2016), while TCRP Report 195 (2018) found no clear relationship between peak-hour TNC activity and long-term changes in public transit usage. The For-Hire Transportation Study (City of New York, 2016) concluded that speed reductions in NYC are driven by increased freight movements, construction, population growth, and a growth in tourism. August 2018 San Francisco Congestion Trends Report

11 THIS PAGE INTENTIONALLY LEFT BLANK San Francisco Congestion Trends Report August

12 LEGEND Study Segments US 101 (Van Ness Ave) US 101 (I-280 to SFO Intl. Airport) I-580 (Oakland to Richmond) SR 24 (Oakland to Walnut Creek) US 101 (Palo Alto to San Jose) Study Bridges Golden Gate Bridge Oakland Bay Bridge San Mateo Bridge Study Areas Figure 1: San Francisco Bay Area Study Map 6 August 2018 San Francisco Congestion Trends Report

13 Methodology The study methodology for this work included identifying data sources, compiling results from applicable data, and assessing the relative magnitude of overall congestion trends and changes in Lyft s customer demand and use of Lyft services. This study relied largely on transportation performance data from INRIX, a global leader for transportation analytics. INRIX released its latest 2017 Traffic Scorecard report in February Each year, the Scorecard provides an analysis that compares the year-to-year state of traffic congestion in countries and major metropolitan areas worldwide. INRIX previously prepared a Congestion Trends Report for greater London, United Kingdom, and Transpo Group leveraged the same methodology for similar studies in New York City and Chicago. STUDY APPROACH We performed five different analyses throughout the Bay Area. These analyses included segment analyses and area-wide analyses. Our area-wide analyses address three central city areas in San Francisco, Oakland and Berkeley, and San Jose. Our segment analyses include a Major Roadway Analysis and Critical Streets Analysis. The Major Roadway Analysis addresses eight significance in the Bay Area, comprising four freeway segments, three bridges, and one arterial route. Supplementing our area-wide analysis of San Francisco east of Van Ness Avenue is the Critical Street Analysis, an examination of three segments of San Francisco city streets. These segments were chosen on the basis of Lyft activity for the purposes of identifying any coincidence with changes in travel speeds; this analysis differs from the Major Roadway Analysis because the street segments were selected on the basis of high Lyft activity as opposed to being selected on the basis of growing congestion on major road links. Figure 2 shows the methodology of the performed analysis to examine congestion and TNC trends. Major Roadway Analysis For the Bay Area Study, we chose to assess key roadway segments that experience the most congestion and three central city areas with the most Lyft activity. Based on an overview of INRIX data, the following roadway highly-congested major roadways of regional San Francisco Congestion Trends Report August

14 Figure 2: Study Methodology segments were selected as key segments in the overall transportation network: US 101 (Van Ness Avenue) US 101 (I-280 to San Francisco International Airport) I-580 (Oakland to Richmond) SR 24 (Oakland to Walnut Creek) US 101 (Palo Alto to San Jose) Golden Gate Bridge (US 101) Oakland Bay Bridge (I-80) San Mateo Bridge (CA 92) Critical Streets Analysis In addition to the segment analysis for major roadways, we also performed a Critical Streets Analysis. This analysis is an examination of areas with the highest Lyft activity in San Francisco and provides a contrast of that activity with congestion. The three selected high-activity streets include the following: Geary Boulevard between Mason St and Leavenworth St Taylor Street between Turk St and California St Sutter Street between Stockton and Larkin St Center City Area Analysis In addition to the segments, we defined a study area comprising the roadway networks within northeast San Francisco, the Oakland/Berkeley area, and downtown San Jose, all displayed in Figure 1. We supplemented the transportation data analysis of the Bay Area with data related to economic activity, population and vehicle registration, bicycling, and public transit. Using this information, independent of any TNC data, we were able to draw conclusions concerning the 8 August 2018 San Francisco Congestion Trends Report

15 changing transportation system in the Bay Area and probable trends related to those changes. Transit and Caltrain Analysis We also analyzed transit ridership throughout the Bay Area and a Caltrain station-specific contrast of ridership activity with Lyft activity serving the immediate areas for the selected high-growth stations. In order to study the effects of TNCs on congestion, we requested data from Lyft. This data was used to identify the extent of Lyft services on the individual roadway segments and the three central city areas. We also obtained data on the high-occupancy Lyft Line service, noting the overall level of Lyft Line services in the Bay Area. Transportation Data Several data sources were used to support the various aspects of this analysis. One of the main sources is transportation performance data from the INRIX network, which includes 300 million vehicles, smart phones, cameras, and other sensors with the ability to cover nearly 5 million miles of roads, ramps and interchanges in over 50 countries. To complement the INRIX Table 1: Congestion Trends Report Data Sources Description Source Traffic Conditions Average Road Speeds, Travel Times INRIX Travel Demand and Capacity Drivers U.S. Gross Domestic Product U.S. Bureau of Economic Analysis Employment Population Gas Prices E-commerce Vehicle Registration Taxi Medallions Modeshare Bicycle Volumes Bicycle network Average Speeds, Travel Times Critical Links Traffic Volumes Bus and Subway Ridership Lyft Trends Performance Trends U.S. Bureau of Labor Statistics U.S. Census Bureau U.S. Energy Information Administration U.S. Census Bureau California Department of Motor Vehicles San Francisco Municipal Transportation Agency San Francisco Municipal Transportation Agency San Francisco Municipal Transportation Agency San Francisco Municipal Transportation Agency INRIX California Department of Transportation Bay Area Rapid Transit, Caltrain, San Francisco Municipal Transportation Agency Lyft San Francisco Congestion Trends Report August

16 data, we utilized information from a variety of recognized authorities within the Bay Area region, including the California Department of Transportation (Caltrans), San Francisco Municipal Transportation Agency (SFMTA), Bay Area Rapid Transit (BART), and Caltrain. All of the data obtained from public agencies was gathered through publicly-available information sources such as web sites and online searches for public information shared by those agencies. Lyft Data Lyft delivered the Bay Area trip data for the first two weeks of October for all the data sets. The month of October is representative of peak demand and peak transportation system performance due to school schedules, holidays, The data obtained from Lyft was aggregated and entirely anonymized prior to being transmitted to the research team. None of the data that Lyft provided to the research team was used to trace individual route choices or obtain information on specific customers. Public Agency Data We ve used the information from a variety of recognized transportation authorities such as Caltrans, BART, Caltrain, and SFMTA. We obtained socio-economic characteristics using data from the U.S. Bureau of Economic Analysis, the U.S. Bureau of Labor Statistics, the U.S. Energy Information Administration, the U.S. Census Bureau, and the California Department of Motor Vehicles. This report s data sources are summarized in Table 1. waning construction activity, and the low likelihood of inclement weather. The data sets include: Pulse (trips in 10 second increments) for the first two weeks in October 2017; Ride data with information about a trip s origin and destination (o/d data), time of passenger pick-up and drop-off, and status of the ride (finished/canceled/ in progress) for ; Aggregated area overview data parsed by day of the week, hourly brackets, all Lyft rides, and Lyft Line rides, for August 2018 San Francisco Congestion Trends Report

17 Travel Demand & Capacity Drivers Transportation is a derived demand, meaning that trips are made as a means of getting someone or something from one location to another, not for the sake of the journey. People travel to work, for business meetings, for social activities, for educational and health reasons, as well as to purchase goods or services required for daily life. Population growth drives travel demand in addition to many other economic factors that result in growing travel activity. In this section, we examine changes in economic activity, population, vehicle registrations, commerce, and travel mode choices. Overall, our research concluded that the Bay Area s population is increasing, freight and associated e-commerce deliveries are increasing, and employment continues to rise within the Bay Area. Long-distance commuting in the Bay Area has increased due to a lack of residential development near employment centers The Bay Area population increased over 2.5 percent throughout the study period. The fraction of employed Bay Area residents has increased to almost 97 percent. Only 3 percent of San Franciscans reported using TNCs daily In San Francisco, the increase in TNC use has only resulted in a decrease in private auto use San Francisco Congestion Trends Report August

18 ECONOMIC AND EMPLOYMENT INDICATORS Gross Domestic Product Increased The nine-county Bay Area has experienced a 16 percent growth in the Gross Domestic Product (GDP) between 2014 and Silicon Valley, a global center for technology and disruptive innovation, is located in the south of the Bay Area area, drives a large portion of GDP growth. Figure 3 shows the GDP growth. More Bay Area Residents are Employed The percentage of Bay Area residents who are employed increased by 3.2 percent, from 94 percent in January 2014 to 97.2 percent in December Figure 4 summarizes these employment trends. Overall, 357,000 more Bay Area residents became employed, a significant driver of mobility demand in the region Figure 3: Gross domestic product (GDP) by metropolitan area ( ) Source: U.S. Bureau of Economic Analysis Dollars (in millions) $900,000 $800,000 $700,000 $600,000 $500,000 $400,000 $300,000 $200,000 $100,000 $ Year San Francisco-Oakland- Hayward, CA (MSA) San Jose-Sunnyvale- Santa Clara, CA (MSA) Santa Rosa, CA (MSA) Vallejo-Fairfield, CA (MSA) Figure 4: San Francisco Bay Counties Employment Rate ( ) Source: U.S. Bureau of Labor Statistics 98% Employment Rate 97% 96% 95% 94% 93% 92% Year Employment Rate 12 August 2018 San Francisco Congestion Trends Report

19 POPULATION AND VEHICLE USE Population Increase Population growth typically involves additional vehicle use, particularly if that growth is occurring in areas not served by frequent or reliable transit. The population of the Bay Area overall increased smaller cities experience a population boost due to commuters. For instance, Palo Alto (located southeast of San Francisco) has a residential population of approximately Bay Area Population 7.6M Source: U.S. Census by more than 2.5 percent during the study period. All study areas experienced population growth. San Francisco population increased by 3.6 percent (more than 30,000 residents). San Jose population increased by 1.8 percent (more than 18,000 residents). Oakland and Berkeley population has increased by more than 2.5 percent or by more than 11,000 and 3,000, respectively. The daytime population of urban areas, typically, is larger than the residential population. San Francisco, San Jose, and Oakland are concentration points for employment, business and cultural activities. According to the U.S. Census Bureau, in 2010, San Francisco's daytime population increases by 21 percent due to commuting. In the Bay Area, because many employers and institutions are 64,000 people, is home to large employers, which results in the city s population increasing by more than 80 percent (51,000) during the day. Housing Due to the Bay Area s booming economy, the demand for housing is high as well. However, the rate of population growth is outpacing the housing growth and is causing housing shortage in the Bay Area. According to a report by Next 10, a non-profit organization, the rate of housing permit issuance is very low for every 100 new residents there are only 24.7 housing permits. This housing undersupply forces people to move further from the place of employment in urban areas, contributing to traffic on major roadways leading to San Francisco and other major employment areas. located outside of the urban areas, some of the San Francisco Congestion Trends Report August

20 Figure 5: 2017 Driver Population and Vehicle Use Source: SFMTA, California DMV SF Taxi Medallions Driver Licenses Registered Vehicles 1,800 (2017) 5,490,520 (2017) 6,701,009 (2017) Gas and Diesel Prices Fell Both gasoline and diesel prices declined from early 2014 to mid However, prices began to rise in mid-2016 and have returned to early 2014 prices by 2018, as shown in Figure 6 More Drivers on the Road According to the California Department of Motor Vehicles (CDMV), the number of driver licenses issued in the 9 counties of the Bay Area increased by 9 percent from 2013 to 2017, an addition of over 450,000 new drivers. A Changing Landscape According to the SFMTA, there are 1,800 taxi medallions in the city of San Francisco. According to CDMV, there were more than 6.7 million of registered vehicles in the Bay Area. Figure 6: Gasoline and Diesel Prices (2014 to 2018) Source: U.S. Energy Information Administration Price Per Gallon (Dollars) $5.00 $4.50 $4.00 $3.50 $3.00 $2.50 $2.00 $1.50 $1.00 $ Year San Francisco California 14 August 2018 San Francisco Congestion Trends Report

21 COMMERCE & FREIGHT While e-commerce creates more delivery truck traffic, some delivery trips are likely to replace individual trips that would have otherwise been made to the grocery store, the bank, or any other While e-commerce data specific to the Bay Area was not available, it is expected that the national trend (shown in Figure 7) is representative of the e-commerce activity within the Bay Area. brick-and-mortar shop. However, these trips still occur and those facilities remain open for business. Figure 7: Quarterly U.S. E-commerce (2014 to 2017) Source: Retail Indicators Branch, U.S. Census Bureau $140,000 U.S. online Sales (millions of dollars) $120,000 $100,000 $80,000 $60,000 $40,000 $20,000 $0 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q Year BAY AREA MODESHARE E-Commerce According to a study conducted by SFMTA about mode share within San Francisco, between 2014 and 2017 the number of people who drove to work (alone or with others) decreased from 46 to 43 percent, and the number of people who took taxi, TNCs, or bicycles increased by three percent. Transit mode splits did not change during this time, while walking increased by one percent. This suggests that within San Francisco, the increase in TNC use has only resulted in a decrease in private auto use as shown in Figure 8. Outside of San Francisco, the mode share is and fewer people use TNCs as their primary mode of travel (shown in Figure 9). The study also posed a question about usage frequency of new travel options such as TNC or Chariot. Only 3 percent of respondents in San Francisco use TNC on daily basis and 23 percent use TNCs rarely. Among non-san Francisco respondents, only 1 percent use TNC daily and 29 percent rarely use a TNC service. These results suggest that TNCs are not a part of people s everyday routine in the city of San Francisco itself. similar, but more people use private automobiles San Francisco Congestion Trends Report August

22 Figure 8: San Francisco Mode Share Change ( ) Source: SMFTA 2014 Private Auto Non-Private Auto 30% Drove Alone 26% Transit 16% Drove with Others 24% Walk 4% Taxi/TNC/Bicycle 2017 Private Auto Non-Private Auto 28% Drove Alone 26% Transit 15% Drove with Others 25% Walk 0% 10% 20% 30% 40% 50% 60% 70% 7% Taxi/TNC/Bicycle Bicycle Operations There are several bikeshare companies operating in the San Francisco area, including both dock-based and dockless bike sharing programs. Ford GoBike share program has 7,000 bicycles across San Francisco, San Jose, Oakland, Berkeley, and Emeryville. In October 2017 there were more than 86,000 total trips with average 18 minutes trip duration in San Francisco. Two dockless bike share companies operate in Bay Area cities Jump and LimeBike. Each year, SFMTA measures bicycle ridership in San Francisco by collecting data from 75 automated bike counters. More than 44,000 bicycles were counted on the average weekday in However, based on available bike ridership data from 24 automated bike counters, cumulative ridership dropped by 4.5 percent from 2016 to 2017 (shown in Figure 10). This trend might be a result of multiple factors, which can include weather, street closures, or growth in non-motorized mobility options. Figure 9: San Francisco & Outside San Francisco Mode Share (2017) Source: SMFTA San Francisco 2% 4% 1% Outside San Francisco.25% 1.5%.25% Drive Alone Carpool 27% 27% 32.5% 30.5% Walk Transit Bicycle 27% 15% 18.5% 16.5% TNC Taxi/Carshare/Other 16 August 2018 San Francisco Congestion Trends Report

23 Figure 10: San Francisco Bicycle Ridership ( ) Source: SMFTA Ridership 450, , , , , , , ,000 50,000 0 Jan Feb March April May June July Aug Sept Oct Nov Dec Month Protected Bike Lanes Figure 11 displays the composition of bike lanes in San Francisco in According to the SFMTA, San Francisco has more than 421 on-street bikeways including 13 miles of protected bicycle lanes and 70 miles of bike trails throughout the city. In 2017, more than 14 miles of bikeways were upgraded or added to the city s network. While it remains outside the scope of this document to evaluate motor vehicle travel pattern changes caused by the removal of travel lanes or the removal of parking lanes, the bicycle network expansion is disruptive of overall capacity and travel demand balancing within the street network, even if users of individual segments may experience only minor changes in travel times or average speeds. Figure 11: San Francisco Bicycle Network (2016) Source: SFMTA Shared Lanes Standard Bikeways Bike Trains Protected Bikeways Miles San Francisco Congestion Trends Report August

24 Ride-Hail Circulation Many studies have examined the effects of taxi circulation in congested cities and have determined that this circulation has an impact on congestion. The circulation of TNC vehicles, however, is typically purposeful and part of the driver s activities in moving to the next pickup location. TNC operators have an interest in reducing vehicle downtime and thereby limit circulation to the extent practicable. 18 August 2018 San Francisco Congestion Trends Report

25 Performance Trends Our approach to analyzing the Bay Area transportation performance trends is multifaceted, utilizing a variety of techniques to ensure a reliable capture of various performance metrics. This section will address public transit trends, major roadway trends (the five major roadways and three bridges with the greatest congestion), critical area trends (the three central city areas), and critical street segment trends (the three San Francisco street segments with the greatest Lyft activity). Transit ridership in the Bay Area has increased by 4 percent during the study period Lyft trips account for less than 1 percent of daily traffic along the Bay Area bridges (San Mateo, Golden Gate and Bay Bridge) Lyft pickups and drop-offs outside San Francisco are concentrated around Caltrain and BART stations 70 percent of Lyft trips start and end in the same city, indicating that Lyft does not add to growing regional congestion Lyft Line trips account for 40 percent of all Lyft trips in the Bay Area San Francisco Congestion Trends Report August

26 PUBLIC TRANSIT TRENDS The Bay Area has an extensive public transportation system that includes: Heavy rail system, BART, that connects San Francisco to the East Bay, and two international airports. BART operates from early morning till midnight on weekends and runs every 15 minutes on weekends and every 20 minutes on weekends and nights. Commuter rail line, Caltrain, that serves San Francisco Peninsula and Santa Clara Valley. Light rail systems, Muni Metro and VTA Light Rail, that serve San Francisco and Santa Clara County accordingly. Bus systems, Muni in Sam Francisco, AC Transit in Alameda and Contra Costa Counties. Ferry services, Golden Gate Ferry that connects San Francisco and Marine County, and San Francisco Bay Ferry that connects San Francisco with Alameda, Contra Costa, Marine and Solano Counties. The following analysis focuses on BART, Caltrain, and Muni transit systems that connect residential areas to different activity centers in the Bay Area. As displayed in Figure 12, the total number average weekday passenger boardings increased by 4 percent from million passengers in 2014 to million of passengers in However, there was a slight ridership decrease in 2017 across the transit systems. Our assessment of the effect of TNC operations on public transit ridership in the Bay Area is supported by analysis concerning the change in public transit ridership and change in Lyft trips in the areas operated by different transit agencies. It is not possible to conclusively demonstrate that a decline in public 20 August 2018 San Francisco Congestion Trends Report

27 Figure 12: Average Weekday Passenger Boardings ( ) Source: Caltrain, SMFTA, BART 1,200,000 1,100,000 1,000,000 Boardings 900, , , , , , Year Muni BART Caltrain transit ridership is consistent with or caused by a growth in Lyft trips. The areas with the biggest Lyft growth during the study period have experienced increases in public transit ridership yet areas with the largest transit ridership decrease did not experience the largest growth of Lyft trips. Transit ridership increases for San Francisco Muni, Caltrain, and San Francisco Ferry systems range from 2 percent to 30 percent. The areas where these particular public transit systems operate experienced the largest increases in Lyft trips in the region. In contrast, Golden Gate Transit, connecting San Francisco with Marin County, experienced the largest decrease in ridership, an approximate 31 percent decrease from 2014 to Using the same regional link, Lyft trips account for less than 1 percent of the vehicle volumes crossing the Golden Gate Marin County was among the smallest of all Bay Area Counties. The same trend of decreasing ridership can be observed with Santa Clara Valley Transportation Authority and the Alameda-Contra Costa Transit District. Those areas observed public transit ridership decreases by 1.2 and 3.9 percent, respectively. These areas experienced greater Lyft growth than Marin County but markedly less than other areas where transit use grew. The recent transit ridership decrease cannot therefore be associated with the Lyft trips growth in the Bay Area, based on the contrast of Lyft service growth with transit ridership increases. Changes to transit ridership may be the result of a variety of economic factors, including increased car ownership, lower fuel prices, and/ or disruptions and unreliability associated with some public transit service options. Bridge. Moreover, the growth of Lyft activity in San Francisco Congestion Trends Report August

28 Figure 13: Segment Speed & Travel Time Changes Source: INRIX, California Department of Transportation MAJOR ROADWAY TRENDS This section summarizes the findings on the four freeway segments, the one arterial segment, and the three major bridge links. Historical data for every Friday (one of the busiest weekdays, and the weekday with peak Lyft activity) between 2014 and 2017 was used to analyze general trends on Bay Area roadway operations. Figure 13 summarizes the change in average daily speeds and volumes for the key roadway segments in the Bay Area from 2014 to Average daily travel speeds on six of the eight study roadways have decreased in the last four years (Oakland Bay Bridge and San Mateo Bridge speeds slightly increased). Roadway speeds decreased between one and seven percent on the six other roadways between 2014 and Study Segment US 101 (Van Ness Ave) US 101 (I-280 to SFO Intl. Airport) I-580 (Oakland to Richmond) SR 24 (Oakland to Walnut Creek) US 101 (Palo Alto to San Jose) Golden Gate Bridge Oakland Bay Bridge San Mateo Bridge Speed/Volume Change -2%/-1% -4%/+7% -7%/+5% -5%/+6% -1%/+1% -1%/+5% +0.3%/+3% +2%/+9% Average daily traffic (ADT) increased on seven of the eight study roadways between 2014 and 2016 (volumes decreased about one percent on Van Ness Ave in San Francisco). The volume increases were between one and nine percent, with the greatest increase occurring on the San Mateo Bridge (+9 percent). CENTRAL CITY AREA TRENDS In addition to using eight key roadway segments as a key indicators of Bay Area roadway Note: Average daily speed change from , and average daily volume change from performance, we also assessed the following 22 August 2018 San Francisco Congestion Trends Report

29 three central city areas: northeast San Francisco, the Oakland/Berkeley area and downtown San the average speeds for each of the study areas on a typical Friday from 2014 to Jose (see Figure 1). Figures summarize Figure 14: Average Speed in San Francisco Study Area ( ) Source: INRIX Average Friday Speeds (mph) am 1am 2am 3am 4am 5am 6am 7am 8am 9am 10am 11am 12pm 1pm 2pm 3pm 4pm 5pm 6pm 7pm 8pm 9pm 10pm 11pm Time Figure 15: Average Speed in San Jose Study Area ( ) Source: INRIX Average Friday Speeds (mph) am 1am 2am 3am 4am 5am 6am 7am 8am 9am 10am 11am 12pm 1pm 2pm 3pm 4pm 5pm 6pm 7pm 8pm 9pm 10pm 11pm Time Figure 16: Average Speed in Oakland Study Area ( ) Source: INRIX Average Friday Speeds (mph) am 1am 2am 3am 4am 5am 6am 7am 8am 9am 10am 11am 12pm 1pm 2pm 3pm 4pm 5pm 6pm 7pm 8pm 9pm 10pm 11pm Time San Francisco Congestion Trends Report August

30 Figure 17: Lyft Trips Growth vs Speed Change in Study Areas ( ) Source: Lyft, INRIX All three study areas experienced a decrease in travel speeds from 2014 to 2017 throughout all hours of the day. The smallest decrease occurred in San Francisco, where average daily speeds decreased by approximately 11 percent. However, average speed has increased by 0.1 miles per hour between 2016 to Average daily travel speeds decreased by 13 and 17 percent during the study period in Oakland/ Berkeley and San Jose, respectively. PM Peak Hour (5 PM) speed change was minimal in San Francisco as well during the study period. PM Peak Hour speed decreased by 9 percent in San Francisco study area, by 22 percent in Oakland study area, and by 24 percent in San Jose study area. Figure 17 summarizes changes in Lyft trips and average speeds over the study period. The areas that experienced the most significant growth in Lyft trips did not see the major average speed decrease. Figure 18: Average Number of Lyft Trips by Time of Day (October 2017) Source: Lyft Average Number of Lyft Trips 14,000 12,000 10,000 8,000 6,000 4,000 2,000 0 Monday Tuesday Wednesday Thursday Friday Saturday Sunday Day Lyft Trips 24 August 2018 San Francisco Congestion Trends Report

31 CHANGING RIDESHARE LANDSCAPE Since its launch in the Bay Area in June 2012, Lyft service has grown each year. Figure 18 summarizes the average number of Lyft trips by time of day. In October 2017, Lyft provided an average of 140,000 trips per day within the Bay Area. Lyft activity is highest on Friday and the weekends and the lowest on Monday. Spatial Analysis of Lyft Trips Additionally, we analyzed the locations of ridesharing pick-ups and drop-offs to identify the areas where the demand is high and where any adverse effects of ride-sharing trips may occur. Figures 19 and 20 provide spatial details of the Lyft s pick-up and drop off locations on an average day in October Trip pick-ups and drop-offs were aggregated to tracts, a commonly used geospatial unit for analysis, and then normalized to be units of pickups per square mile. The majority of Lyft s pick-ups and drop-offs occur in northeast (downtown) San Francisco, where there are as many as 20,000 pick-ups and dropoffs per square mile on an average day. Outside of San Francisco the areas of highest pick-ups and drop-offs tend to occur around Caltrain and Bart stations suggesting that Lyft is serving as a first/last mile connection to existing high occupancy transit. This seems to make sense, Table 2: Average Number of Daily Pickups and Drop-Offs near Caltrain Stations (October 2014 & 2017) Source: Lyft Station 2014 Average Pickups & Drop-Offs 2017 Average Pickups & Drop-Offs Palo Alto Redwood City 3 64 Sunnyvale Mountain View Station San Jose Diridon Milbrae Hillsdale 7 80 San Mateo 3 25 Menlo Park 4 57 over the last 4 years transit mode split has remained the same while private automobile use has decreased as TNC use has increased. This suggests that Bay Area travelers who previously traveled to and from BART and Caltrain stations by private automobile may now be using TNCs to do so, which saves the cost and hassle of parking. Lyft activity is also high around SFO. Caltrain Stations: First/Last Mile The transit and Lyft interaction was also evident at Caltrain stations. Our analysis of Caltrain stations with the biggest train ridership growth during the study area indicated that those large increases were typically accompanied by growth in Lyft trips serving the station area. These trips are defined as Lyft rides starting or ending either at the Caltrain station or in the Caltrain station parking lot, using average weekday Lyft and Caltrain ridership data. San Francisco Congestion Trends Report August

32 LEGEND Average Daily Pickups Per Square Mile < ,300 1,300-2,200 2,200-3,600 3,600-5,200 5,200-9,000 9,000-21,475 BART Route Caltrain & ACE Routes Figure 19: Average Daily Lyft Pickups in October August 2018 San Francisco Congestion Trends Report

33 LEGEND Average Daily Drop- Offs Per Square Mile < ,300 1,300-2,200 2,200-3,600 3,600-5,200 5,200-9,000 9,000-21,475 BART Route Caltrain & ACE Routes Figure 20: Average Daily Lyft Drop-Offs in October 2017 San Francisco Congestion Trends Report August

34 Table 3: Top Bay Area Lyft Origins and Destinations Outside San Francisco (October 2017) Source: Lyft Origin Destination Daily Average Lyft Trips Oakland Oakland 7,356 San Jose San Jose 4,911 San Francisco San Bruno 2,998 San Bruno San Francisco 2,551 Berkeley Berkeley 2,329 San Francisco Oakland 1,115 Oakland Berkeley 1,089 Berkeley Oakland 1,054 Oakland San Francisco 860 Table 2 shows the stations analyzed and average number of pickups/ drop offs during average weekday in October (Tuesday-Thursday) in 2014 and Lyft appears to be serving more station-area trips, reducing the need for additional parking facilities in the region and possibly reducing the need for vehicle ownership in the Caltrain corridor. Lyft Trip Flows To analyze the trip flows between areas in the Bay Area, trips origin and destination were matched and aggregated by urban areas. Table 3 shows the most popular origins and destinations outside of San Francisco in October The overall trend is that more than 60 percent of the Lyft s trips start and end within the urban areas like San Francisco, Oakland or San Jose. San Francisco Speeds and Lyft Activity In this section, we assess the performance of street segments in the areas of greatest Lyft activity. After examining Lyft data, we selected three roadway segments that experience both a high rate of Lyft activity and traffic demand related to commercial and retail uses. The assessment of average roadway speeds during the day and Table 4: Friday Lyft Pickups & Drop-Offs (October 2017) Source: Lyft Time Geary Boulevard Between Mason & Leavenworth Taylor Street Between Turk & California Sutter Street Between Stockton & Larkin Pickups & Drop-Offs % Total Trips Pickups & Drop-Offs % Total Trips Pickups & Drop-Offs % Total Trips 12am-3am 56 8% 47 8% 68 5% 4am-7am 33 5% 34 6% 85 6% 8am-11am 95 14% 94 15% % 12pm-3pm 78 12% 67 11% % 4pm-7pm % % % 8pm-11pm % % % 28 August 2018 San Francisco Congestion Trends Report

35 hourly Lyft activity are based on Fridays in October 2017 as Fridays exhibit the heaviest Lyft activity during the week. Table 4 shows the road segments chosen and the time distribution of Lyft pickups and drop-offs. The identification of areas with high Lyft activity allowed us to identify critical street segments. The speeds for an average Friday were determined using INRIX data and the results were used to create a Figure 21: Average Speed and Lyft Activity on Geary Blvd Between Mason St and Leavenworth St, Fridays in October 2017 Source: INRIX, Lyft Number of Pickups & Drop-Offs :00am-3:00am 4:00am-7:00am 8:00am-11:00am 12:00pm-3:00pm 4:00pm-7:00pm 8:00pm-11:00pm Time Average Speed (mph) Figure 22: Average Speed and Lyft Activity on Taylor St Between Turk St and California St, Fridays in October 2017 Source: INRIX, Lyft Number of Pickups & Drop-Offs :00am-3:00am 4:00am-7:00am 8:00am-11:00am 12:00pm-3:00pm 4:00pm-7:00pm 8:00pm-11:00pm Time Average Speed (mph) Figure 23: Average Speed and Lyft Activity on Sutter St Between Stockton St and Larkin St, Fridays in October 2017 Source: INRIX, Lyft Number of Pickups & Drop-Offs :00am-3:00am 4:00am-7:00am 8:00am-11:00am 12:00pm-3:00pm 4:00pm-7:00pm 8:00pm-11:00pm Time Average Speed (mph) Lyft Pickups & Drop-Offs Average Speeds San Francisco Congestion Trends Report August

36 time-spread speed distribution, summarized in Figures 21 through 23, along with the Lyft activity for each of the selected road segments. The majority of the Lyft pickups and drop-offs on Geary Blvd and Taylor St occur after 8 PM. On Sutter St, the peak Lyft activity starts around 4 PM and remains elevated after 8 PM. This activity is consistent with activity in similar commercial areas. The spike in early-evening Lyft activity may be the result of changing transit route headways or service terminations that occur at 7 PM, reducing transit service during the time period following the evening commute hours. Those desiring to use transit are left with few options and likely comprise some of the Lyft activity in the early evening hours. All three road segments experienced either AM or PM peak hour when the speeds are the lowest during the day. The analysis for all three of these road segments reveals that Lyft s busiest hours are outside of the time when the speeds are the lowest. On Geary Blvd, the lowest speeds occur from 4 to 7 PM, but the heaviest Lyft activity occurs between 8 PM and 11 PM. On Taylor St, the hours exhibiting the lowest travel speeds are 8 AM to 11 AM and 4 PM to 7 PM; this is contract to the hours exhibiting the busiest Lyft activity, between 8 PM 11PM, when the segment speeds are increasing. The same trend is visible on Sutter Street, where the lowest speeds occur during morning peak hour between 8 AM and 11 AM but the Lyft activity is the heaviest between 4 PM and 7 PM and that period of activity extends beyond 8 PM when segment speeds are increasing. 30 August 2018 San Francisco Congestion Trends Report

37 HIGH-OCCUPANCY RIDESHARE TRENDS Lyft introduced Lyft Line service in the Bay Area in mid This service allows riders to join trips in progress, creating a high-occupancy service that is not provided by taxis. As displayed in Figure 24, Lyft Line trips continue to grow and presently comprise approximately 40 percent of all Lyft trips in the Bay Area. Lyft s increasing multi-passenger trips are aiding fraction of vehicles that are high-occupancy and eliminating the need to store cars in the city core. As more people abandon single-occupant taxis and personal vehicles in favor of convenient Lyft Line service, Lyft s ability to change the ratio of high-occupancy to single-occupancy vehicles is likely to make an even bigger impact on carbon emissions and reduced personal vehicle trips. travel throughout the Bay Area by increasing the Figure 24: Lyft Line Trips in San Francisco Bay Area ( ) Source: Lyft Lyft Trips/Month 5,000,000 4,500,000 4,000,000 3,500,000 3,000,000 2,500,000 2,000,000 1,500,000 1,000, ,000 0 Jan Mar May Jul Sep Nov Jan Mar May Jul Sep Nov Jan Mar May Jul Sep Nov Jan Mar May Jul Sep Nov Date Total Lyft Trips Lyft Line Trips San Francisco Congestion Trends Report August

38 THIS PAGE INTENTIONALLY LEFT BLANK 32 August 2018 San Francisco Congestion Trends Report

39 Findings & Conclusions This study explored the drivers of congestion and evaluated various aspects of transportation system performance in the Bay Area. Transpo Group identified various publicly-available data sources related to economic growth, population and demographic trends, and transportation system management. After evaluating the drivers of demand, we obtained Lyft data in order to understand how Lyft serves the Bay Area trips. Our evaluation of the congestion drivers revealed that trip growth is Regional changes such as population and employment growth, as well as the lack of housing near employment centers likely play larger roles in the increase in congestion. being driven by economic activity, justified by the following findings: The population of the Bay Area is increasing along with the fraction of employed people in the region Housing shortages in areas of employment growth result in people living and commuting from areas away from their place of employment, contributing to increased regional congestion Approximately 70 percent of Lyft trips in the Bay Area are intracity trips, meaning they start and end within the same city and do not add to the growing congestion on regional roadways. Lyft trips on the three study bridges represent less than 1 percent of average daily roadway volumes. High-occupancy Lyft trips comprise 40 percent of all Lyft trips in the Bay Area. San Francisco Congestion Trends Report August

40 Bay Area locations with the largest growth in Lyft activity have not had the largest decrease in average daily speeds. Areas with little Lyft growth have seen more dramatic increases in congestion, suggesting that Lyft may not be the primary factor in congestion growth in the Bay Area. It is not possible to conclusively demonstrate that a decline in public transit ridership is consistent with or caused by a growth in Lyft trips. The areas with the biggest Lyft growth during the study period have experienced increases in public transit ridership yet areas with the largest transit ridership decrease did not experience the largest growth of Lyft trips. The Bay Area s expanding economy, along with population and employment growth, resulted in a travel demand increase in the area. Relatively inexpensive fuel prices and the lack of housing availability near employment centers have all contributed to the growth of congestion in the Bay Area. Travel speed reduction on the study road segments, often understood as an indicator of congestion, is not concomitant with growth in Lyft trips. The observations in Figures 15 and 16 seem to indicate that the majority of Lyft trips are concentrated within cities and those trips do not add to congestion on regional major highways. Moreover, a recent travel decision survey conducted by SFMTA showed that only 4 percent of people in San Francisco and 2 percent in the greater Bay Area use TNCs to commute to work. Lyft trips also account for less than 1 percent of daily traffic on the San Mateo, Golden Gate, and Bay Bridges. Travel speeds in the urban study areas have decreased by 7.2 percent in Oakland/Berkeley, 8.6 percent in San Jose, and 4.2 percent in San Francisco during the study period. However, Lyft trip growth was the largest in San Francisco where the 34 August 2018 San Francisco Congestion Trends Report

41 daily speed reduction was the smallest. Moreover, the daily average travel speeds in the San Francisco study area remained the same from 2016 to 2017, despite an average increase of 18,000 pickups per day in that area. This might suggest that the growth in Lyft trips are not the driving factor in the growth of congestion when speed is a consideration. A complex array of factors influences the growth and severity of congestion. Cities around the world have sought to mitigate congestion with a variety of approaches beyond civil infrastructure projects. These strategies include the following approaches: Signal Optimization and Coordination Transit Signal Priority and Adaptive Traffic Signal Control Curb Management and Responsive Freight Management Policies Dynamic Parking Pricing, Monitoring, and Information Systems Camera Enforcement of Loading Zones and Double Parking Hot Spots Congestion Pricing for both Segments and Cordons San Francisco Congestion Trends Report August

42 SUPPORTING DOCUMENTS California Department of Transportation (2014). San Francisco Bay Area Freight Mobility Study Clewlow, R.R. and Mishra, G.S., Disruptive transportation: The adoption, utilization, and impacts of ride-hailing in the United States. University of California, Davis, Institute of Transportation Studies, Davis, CA, Research Report UCD-ITS-RR Li, Z., Hong, Y. and Zhang, Z., An empirical analysis of on-demand ride sharing and traffic congestion. Thirty Seventh International Conference on Information Systems, Dublin. MAPC, Fare choices a survey of ride-hailing passengers in metro Boston. A Metropolitan Area Planning Council Research Brief. McKenzie, B., Koerber, W., Fields, A., Benetsky, M., & Rapino, M. (2013). Commuter-adjusted population estimates: ACS Washington, DC: Journey to Work and Migration Statistics Branch, US Census Bureau. Google Scholar. National Academies of Sciences, Engineering, and Medicine Broadening Understanding of the Interplay Between Public Transit, Shared Mobility, and Personal Automobiles. Washington, DC: The National Academies Press. Next 10 (2018). Growth Amid Dysfunction: California Migration, Current State of California Housing Market, and California Employment by Income Schaller, B., Unsustainable? The growth of app-based ride services and traffic, travel and the future of New York City. Report by Schaller Consulting, Brooklyn NY. SFMTA (2017). TNCs Today. A Profile of San Francisco Transportation Network Company Activity SFMTA (2017). Travel Decision Survey Data Analysis & Comparison Report SFMTA (2018). Pedaling Forward A Glance at the SFMTA s Bike Program for August 2018 San Francisco Congestion Trends Report

43 San Francisco Congestion Trends Report August 2018 THIS PAGE INTENTIONALLY LEFT BLANK 37 37

44