Six-Minute Service
This countdown clock shows a 26-minute gap in trains very close to the evening peak. There were no service alerts at the time of this photo. [1]
Riders Alliance and a coalition of independent activists within the New York region have spent the last year advocating for improvement in off-peak bus and subway frequency. This campaign, dubbed Six-Minute Service, calls for all NYC Transit subway lines and a large number of bus routes to run every six minutes, every day, from morning to evening. The campaign has drawn positive attention from some legislators, who are interested in improving operations. The Effective Transit Alliance wholeheartedly endorses this goal, and believes the benefits of Six-Minute Service have in fact been understated: on the subway, it is most likely a net revenue generator, paying for itself via increased ridership.
Frequency influences passengers' choices about whether to ride transit at all. Shorter waits encourage more passengers to ride, either choosing transit over other modes or making trips they otherwise would not. Long waits have the opposite effect: 10- and 12-minute off-peak and weekend headways discourage ridership, especially for passengers who need to transfer between lines and incur those waits multiple times. When a transit agency responds to budget shortfalls by cutting service, travelers stop using the system, further reducing revenue and exacerbating the original budget shortfalls in a phenomenon commonly called a transit death spiral. As pandemic-related federal operations funding runs out, the MTA must do everything it can to attract riders to the system. This is especially important on the subway, where the fixed costs are high and the savings from service cuts are very small.
Six-minute service also simplifies schedule planning, which improves reliability and thus the rider experience. Today's service planning approach manages each subway route individually based on crowding guidelines, even when those routes share tracks. This strategy leads to hastily-written timetables with uneven headways that trains cannot reliably meet, taxing planning and service management resources. Running all trains on the same frequent headway would permit planners to write achievable schedules.
We cannot fully quantify the benefits of greater reliability, but they are substantial. Passengers dread announcements like “we are being held by the train’s dispatcher” and “we are delayed because of train traffic ahead of us.” In particular, working-class riders who face zero-tolerance policies at work often budget more than half an hour of excess travel time to account for long waits, missed transfers, and train delays. Such riders are using the system now, but are likely to buy a car as soon as they can afford one to escape unreliable commutes that take as much as 90 minutes each way.
Robust academic literature, outlined below, quantifies the benefits of shortening waits. The Comptroller’s office expects six-minute service on weekdays and weekends to require an additional $350 million in annual operating funds. Direct subway costs are a minority of this figure; the rest would go to bus service and maintenance. As we explain below in the section on finances, we expect that the extra ridership and revenue should cover this additional spending, for no net increase in state subsidy. While investments whose costs are easier to estimate than benefits can be daunting, a willingness to take risks is crucial, especially since we can be highly assured that in the medium and long run, efforts like this will more than pay for themselves. If the MTA and the state invest in regular six-minute frequencies, the transit death spiral will be arrested, ridership will increase, and people will keep choosing to forgo car ownership.
The benefits of frequency to ridership
Trains and buses in New York suffer from poor frequency and reliability. Workers know this and budget extra time for their trips, especially if they transfer. The New York Times regularly covers the commute pain of people who work far from where they live. Many workers making two-transfer trips, budget an hour and a half due to 10- and 12-minute waits at each transfer point. Many early shift workers must wake up at five a.m. Unsurprisingly, such workers either move closer for extra sleep or get a car as soon as their finances permit.
Six-minute service would counteract the trend of getting a car to avoid onerous commutes. Regularly spaced trains would bring shorter and more predictable waits than the current system. Riders get this, but the current New York City Transit (NYCT) planning regime does not. Why is that?
The answer has to do with history. NYCT ran trains every four minutes during rush hour as far back as 1968, but the mentality at the time was to use frequency to manage crowding. At very high frequency, this makes sense: the difference to the passenger between running every three minutes and running every four or even five is small, so the only reason to run more trains is if the trains are too crowded. However, at today’s moderate 10-12-minute off-peak frequencies, improving to six-minute frequency would cut wait times enough to persuade more to use the system, particularly for trips outside of traditional peak work commuting hours.
The research literature in public transportation has studied dozens of cases to estimate the effect of frequency on ridership. If frequency is raised by 1%, and ridership rises by x%, it is said that the elasticity of ridership with respect to service is x. A review in Totten-Levinson (2016) suggests that the elasticity of ridership with respect to frequency for frequent urban buses is 0.4. This means that we would expect an increase in off-peak frequency from every 10 minutes to every six minutes—a 67% off-peak service increase—to produce a 23% increase in total off-peak ridership, not simply the same ridership spread out over less-crowded vehicles.
We can also express elasticity in terms of total trip time, including in-vehicle time, wait time, and walking time. The average unlinked subway trip (that is, with each leg of a trip involving a transfer counted separately) is 13 minutes. A study undertaken by Lago-Mayworm-McEnroe (1981) suggests the elasticity of ridership with respect to the combined trip time and worst-case wait time is about -0.7 to -0.8. Thus, the average trip takes 23 minutes at today’s typical off-peak frequencies and would take 19 minutes under six-minute service (a 21% improvement), increasing ridership by 14% to 16%.
Passengers making transfers would reap some of the greatest rewards. The subway system is large enough that trips commonly require two transfers, sometimes with legs of just one or two stations. For riders making these trips, the combined frequency of three minutes produced by two services that interline is game-changing. Such convenient, high-frequency transfers will increase the utility of Governor Hochul’s already transformative Interborough Express (IBX) project. As an orbital line through Brooklyn and Queens connecting to most subway lines, most IBX riders would likely take short trips on the line as part of a two- or three-seat ride. Passengers are much likelier to use this line if the worst-case wait is six minutes rather than 12. In the latter case, riders would often be spending more time waiting than on a train.
Conversely, riders who board trains at outlying stations are less impacted by lower frequency since they make longer trips. For this reason, the A’s Far Rockaway and Lefferts branches are to run every 12 minutes each—still an improvement over today’s service, which is nominally every 15 minutes on weekday afternoons and every 20 minutes all weekend. Six-minute service would start at Rockaway Boulevard station, where the Lefferts Boulevard and Rockaway branches merge.
Subway riders experience long waits during off-peak periods, such as this 11-minute wait at 21st St–Queensbridge. This discourages ridership. [2]
The benefits of frequency to finances
The added fare revenue from six-minute service is likely to cover the fiscal costs of the program, even without taking into account the benefits of better connectivity and modal shift away from cars. This is for two reasons: the costs of the program are likely to be limited, and the benefits of higher frequency to revenue are substantial.
We believe that the Comptroller’s office’s estimate of $350 million in extra annual operating costs is a high-end figure. Most of the cost of subway (but not bus) operations—rolling stock and maintenance, for example—is fixed regardless of off-peak frequency.
Even the marginal crew cost of an off-peak service increase is limited. Peaky schedules require scheduling workers for discontinuous shifts—which workers dislike and demand higher hourly compensation for—or paying for substantial time not spent carrying passengers. Today, NYCT train operators running peaky schedules average 550 revenue-hours per year. By contrast, their counterparts on the London Underground average 720 under less peaky schedules, and those on the Berlin U-Bahn and Helsinki Metro average over 850 running flat five-minute schedules—even as overseas workers generally enjoy more vacation time than Americans.
Moreover, the additional ridership coming from better service translates to added revenue. Before the COVID-19 crisis, about two-thirds of subway ridership was off-peak. While data from this year remains unpublished, the decline in subway ridership has been greater in the peak than off-peak. The 15% rise in off-peak ridership we expect from six-minute off-peak service means overall ridership should grow at least 10%. We cannot be certain how this growth in ridership translates to growth in revenue, since much of the added trips would come from riders who already hold monthly passes. Still, 10% of pre-COVID-19 subway and bus fare receipts is $440 million a year, more than the Comptroller’s projected cost of the service increase.
The benefits of frequency to planning
In addition to the increase in raw capacity, all-day six-minute service also simplifies service planning. This will likely significantly improve reliability, which is crucial for passenger satisfaction. In this section, we explain how this is so.
Already, planners have to plan carefully around merges and service disruptions for maintenance, called general orders or GOs, which usually combine a service change with a slow order near the work zone. Periodic capacity-related tweaks to individual services’ frequencies further complicate matters for planners and passengers; these would disappear under six-minute service.
Today, the subway system suffers from uneven headways due to planning principles that look only at peak capacity, combined with the high degree of track sharing. For example, even though the 2 and 3 trains run together most of the way, the scheduled gaps between trains range from two to seven minutes in the late morning. The E and F trains run every 12 minutes on weekends, but where they run together for a long distance on the Queens Boulevard express tracks, gaps alternate between two and 10 minutes.
In New York, distinct services share tracks to an unusual extent. For example, the D train shares tracks with the A under Central Park West, then with the B through the Manhattan core, and finally with the N in Brooklyn, a practice called interlining or reverse-branching. Each additional interline introduces merges that require careful planning and slower train operation. These merges complicate scheduling even further where two interlined services run on different intervals, such as the shared F and G section between Church Ave. and Bergen St. on which every fifth F train on weekends has to be delayed to allow a G train to turn.
Moreover, the micromanagement of frequency—applying peak-hour planning principles to the entire day—adds work for planning and service management staff. Currently, planners manually adjust the frequency for each individual line based on train crowding data every six months, often with a different schedule for each hour of the day. The resulting timetables contain different, conflicting frequencies for many services on shared track segments. Moreover, they require many trains to go into service just before peak times and go out of service just after—at which point staff have to ensure the train is clear of passengers.
Even with around 300 employees, the planning department is still understaffed compared to present-day scheduling needs. With limited funding for service planning, something has to give. Simplifying service to a largely flat frequency without recurring capacity tweaks would allow planners to refocus resources on making the timetable reliable.
Barriers to six-minute service
There are some technical issues involved with raising off-peak frequencies. On net, six-minute service should make the system easier to plan and operate, but there are some barriers, especially with maintenance. The Comptroller even assumes that these barriers would only allow six-minute frequencies on weekdays, which would incur higher costs of maintenance due to shunting work to the weekend. Fortunately, we have grounds to believe the Comptroller and Riders Alliance are too pessimistic, and in fact these issues are resolvable, permitting high-frequency subway and bus service all day, every day of the week.
Subway maintenance practices
The subway needs regular downtime for track maintenance. In most places in the world, this is done overnight, when trains do not run. But New York runs trains 24/7, and therefore has to do maintenance around service. This is typically done on weekends, but is also sometimes done on weekdays in the off-peak—this is the origin of the constant GOs as mentioned above under planning. To some extent, this practice limits off-peak frequency, and as such the Comptroller’s office believes it is impossible to run six-minute service on both weekdays and weekends.
We believe this is a serious but resolvable issue. While the subway has always run 24/7, the extent of service changes during weekend GOs is recent. In the 2000s, until the service cuts of the Great Recession, many lines ran a train every eight minutes off-peak and on weekends, rather than the 10- and 12-minute headways of today.
Policy changes sold under the guise of safety have expanded the extent of service restrictions next to a work zone track, called flagging. These changes caused a noticeable slowdown in service over the 2010s that was covered in local and national media, including the Daily News, the New York Times, and Vox. Andy Byford reversed some of these slowdowns under the Save Safe Seconds program, but much more can be done.
While New York’s 24/7 service is nearly globally unique, the maintenance challenges are not. In Tokyo, maintenance is done on urban rail lines in three to five hours of nighttime closure, then trains run every five minutes off-peak until midnight again. Japan is where the best practices in the world are found, where there are no GOs at all, but we can learn from Europe as well. Berlin has weekend service changes (albeit fewer than New York, so nearly all lines maintain five-minute service). Paris does intense shutdowns in summertime periods when many Parisians travel elsewhere for vacation, using other lines for redundancy. All of these systems achieve much lower maintenance costs than New York despite confining all or almost all work to short overnight windows. Compare these practices to FasTrack or the recent J/Z shutdown in New York, and you can see that we have appropriate paradigms at our disposal to emulate their successes.
Subway rolling stock
Fortunately, rush hour service on the 1, 2, 3, 4, 5, 6, and 7 lines already meet the six-minute standard; all except the 1 were overcrowded before COVID, so frequency has been set based on capacity constraints. To straighten the schedules would still require a small increase in total vehicles in service, but this can be done with the current fleet. The percentage of trains available for peak service would need to rise from the present-day 85% to about 90%, which peer systems like the London Underground achieve.
Certain lettered lines (the C, D, G, N, Q, R, and W) currently run a maximum frequency worse than every six minutes, so even peak service has to be increased. To bump service up to the six-minute standard at rush hour would require more equipment than is currently available. Fortunately, the fleet will grow in size as new R211 trains enter service starting next year. The combination of new cars and achieving 90% car availability would permit six-minute service.
Road congestion and bus delays
Bus riders are familiar with the experience of waiting 20 minutes for a bus that nominally runs every 10 minutes, and then seeing a second bus arrive immediately after. This phenomenon is called bus bunching. If a bus is slightly delayed, for example due to congestion, then more passengers will be waiting at subsequent stops by the time the bus arrives. This makes each stop take longer, which delays the bus further; eventually, the bus behind it catches up.
To help six-minute bus service run reliably and speedily, the New York City Department of Transportation needs to step in to implement and enforce dedicated bus lanes on all major routes, and give buses signal priority at intersections so that they spend less time waiting at red lights. Moreover, the MTA should implement prepaid fare collection and all-door boarding system wide, rather than just on SBS routes, to cut time spent at stops. San Francisco has done so, and New York City Transit had plans to do the same when Andy Byford was in charge. Finally, it’s essential that fare inspections happen while the bus is moving or riders are at stations, in contrast with current practice on SBS where buses are stopped for inspections.
Three Bx6 buses bunch on Broadway between 155th and 157th Streets due to heavy traffic. [3]
Conclusion
While requiring significant changes to scheduling and maintenance practices, six-minute all-day service on subways and buses would dramatically improve transit’s usability for a vast group of current and potential riders and positively affect NYCT’s finances and operations. The increase in fare revenue from higher ridership can be high enough to offset any additional operating costs, inclusive of new bus and subway car orders, and thus improve the MTA’s fiscal position. As overall transit ridership patterns in New York City and across the States trend away from peak-oriented work commuting and towards a wider variety of trips all day including evenings and weekends, regular frequent service intervals will enable the MTA to recover from its pandemic-induced ridership shock sooner.
The recommendations proposed here are intended to be the starting point of broader systemic investments in subway and bus services, focusing on organization before electronics and concrete, as common to effective transit planning and operations around the world. The faster that ridership and thus fare revenue recover, the sooner upgrades and expansion can be implemented, and ridership gains from increased off-peak service will help inform capacity additions to be included in the 2025-2029 capital budget. Instead of managing stasis or decline as happened in response to the 2008 financial crisis, it is time to refocus on leveraging the City’s transit assets on hand towards overarching goals of financial productivity, socio-economic equity, and climate resilience this decade and beyond.
Footnotes
Photo copyright Jesse Lang, used with permission.
Photo copyright Alon Levy, used with permission.
Photo copyright Alex Sramek, used with permission.
