The Boom in Subway Ridership Is Waning. Why?


Transit officials recently reported that 1,763,000,000 subway trips were taken last year, the most since 1948. But the rise in ridership was meager, with only 12 million more trips in 2015 than in 2014. The percentage growth rate was seven-tenths of one percent. Over the same year, employment in New York City rose three times as fast.

Subway ridership boomed over the prior two decades, rising from 1993 through 2014 at an average annual rate of 34 million trips. During 2013 and 2014 the rate of increase was even higher, averaging 48 million a year, or four times last year’s increase. Measured against those benchmarks, ridership last year fell short by around 30 million trips. At the average effective fare of two bucks, which takes into account price breaks like unlimited farecards, transfers, and senior discounts, NYC Transit lost out on $60 million in revenue in 2015 — no small sum.

Employment in the five boroughs rose last year by 87,000 jobs, or 2.1 percent, according to the Bureau of Labor Statistics. If half of the additional workers rode the subway to work, their daily commutes would have added nearly 22 million trips to the system. Alternatively, if subway ridership had simply risen as fast as employment, the gain in ridership last year would have been 36 million.

But actual trips rose by only 12 million. Why so little? Transit officials won’t comment until “official” 2015 figures are in. But here are some factors that, in combination, could have suppressed subway use last year.

The fare hike. Transit fares went up last March — by 10 percent for a single ride, but much less on a trip-weighted basis since 30-day unlimited fares rose just 4 percent. Subway ridership is notoriously price-inelastic, however, and by my calculations the fare increase would have dampened the number of subway trips last year by a modest three-quarters of one percent, or 10 to 15 million. The next biennial fare hike a year from now will likely have a similar effect.

Cheap gas. Pump prices plunged last year; the nearly 30 percent drop was the steepest in at least a half-century, and helped touch off a 3 percent rise in U.S. driving, the biggest in decades. Though I don’t have data, driving presumably increased in our area, with some of the increase coming at transit’s expense. Even if driving here went up only half as much as nationally, and if just half of that increase came at the expense of transit, the hit to 2015 subway ridership would have been another 10 to 15 million trips.

Uber and other on-demand car hires. Uber’s net congestion impacts in the heart of Manhattan may be an unsettled question, but no one disputes that in the rest of the city app-based for-hire vehicles have taken root faster than traditional yellow taxis have lost ground. A graphic published in FiveThirtyEight in December suggests that last year Uber alone was adding a net of 40,000 or more daily pickups by for-hire vehicles outside the Manhattan core. That would translate to 15 million additional vehicle trips last year, many of which might have substituted for the subway.

However, the most worrisome factor in the slow growth of subway ridership, and the hardest to quantify, is crowding. Jammed trains and platforms aren’t just making subway travel more uncomfortable. They’re forcing passengers into even closer mutual proximity, enabling misogynist and other antisocial behavior and augmenting the sense of disorder that many New Yorkers associate with the subways. And of course crowding slows riders’ entrance and egress from cars and platforms, increasing station “dwell time” and creating delays that then cascade down the line, sometimes for hours. All of these issues would tend to dampen ridership gains.

Internal MTA data reported to its New York City Transit and Bus Committee show only a 1.3 percentage point increase in late subway arrivals last year, from 21.3 percent in 2014 to 22.6 percent (see p. 19). Yet even last June MTA CEO Tom Prendergast was insisting that “crowding and ridership” caused 40 percent of subway delays in 2014 (see p. 21). Moreover, in both 2014 and 2015, provision of train service by the MTA rose only one-third as fast as ridership (thanks to Steve Bauman for the service figures, which employ a standard transit metric called “train stops”), effectively guaranteeing more crowding and more delays.

It may take another year of flattening ridership before we can say for sure that the subways have stumbled into Yogi Berra territory, where “nobody goes there anymore, it’s too crowded.” What’s already clear is that what Prendergast was calling “the renaissance of the New York City subway… a miracle” just a year ago is now, according to New York Magazine, “a delay crisis.”

Prendergast has finally started to speak out about the unfunded MTA capital plan, telling the MTA board last week that the unprecedented delay, which threatens to halt billions in transit upgrades by mid-year, “is wearing on him,” according to Politico. If the MTA CEO, who serves at the pleasure of the governor, is at last finding his voice, then it’s time for state legislators to join Robert Rodriguez and his 14 Assembly co-sponsors in backing a toll-financing plan that could fill the state funding gap and finance the enhanced subway service needed to enable ridership to continue growing.

Filed Under: MTA

  • Alexander Vucelic

    cycling Might be another factor, and that’s a good Thing.

    Citibike daily trips are 40-50,000. Using Citibike as a proxy for Overall cycling One can estimate 200,000 to 250,000 daily cycle trips.

    Walking Might also be another factor. Complete streets has dramaticaly improved walkabilty of Core areas. Even a tiny increase in walk share of trips would Impact subway ridership growth. Manhattan had a commute walk more of 20%. The outer boros roughly 10%. Paris has a 60% walk mode share.

    Finally, the fare increase may be a bigger factor than many of us realize. $5 for a round trip vs. $5.50 might be a deal killer for many trips. I know my subway riding had dropped to essentially zero because of the increase. I know my teenager and his buddies also are making different mobility choices on weekends than they used to.

    Remind me why diversifying mobility choices is a bad thing ?

  • Ian

    Another factor is simply closure of the system. In 2012, Sandy shut down service for three weekdays (~15m trips) and affected it negatively for weeks after. You can see the blip there. In 2015 service was shut down for another weekday for snow. Doesn’t account for all the difference but some of it.

    Comparing average weekday ridership year-to-year is a better metric.

  • com63


  • Is it reasonable to expect a constant smooth increase? That graph shows a wide variety of yearly changes, and even two years in which there were declines from the previous year.

    I don’t see that there is any phenomenon here that needs explaining. If we decided to use time units of six months, the graph would look different. If we used units of 18 months, it would look yet another way. Maybe in one of those graphs the gain in the most recent interval would be greater than in the one before it.

    It seems to me that what is meaningful is the long-term trend, not the short-term fluctuations.

  • Joe R.

    I agree. As an engineer I’ve looked at tons of data in my lifetime. What I see there appears to be statistical noise. There are all sorts of reasons for those variations, such as weather, fewer/more people staying home from work due to illness, and so forth. It might be interesting to see if the dips correlate with years which had lots of inclement weather, or a worse than usual flu season.

  • Joe R.

    I seldom use the subway but the fare certainly affects my decision. I used to go to Manhattan a bit more back when the fares were $1 or under. Ever since they skyrocketed I’ve cut that back considerably. $2.75 each way seems like an awful lot. If people’s salaries had kept pace it would have been different but they haven’t. I know people who were making $10 an hour when the fare was $1 and they’re still making $10 an hour. Quite a few people haven’t had a raise in two decades.

  • ahwr

    Manhattan had a commute walk more of 20%. The outer boros roughly 10%.

    Why are you talking about commute trips here? They are a minority of subway trips. NYMTC’s travel survey gave walk/ bike mode share for Manhattan residents at ~56%, 32% in the rest of the city. Breakdown of the walk/bike split for the NYMTC region was 96% walking, 4% biking. So a small percentage increase in walking has a bigger impact on subway travel than you make out.

    Paris has a 60% walk mode share.

    Are you getting this from wikipedia? If you look at the cite for that stat on page 18 you can see that the number is wrong.

    Walking commute mode share is 10% in the Paris region. I don’t know where wikipedia or you get a 60% walking mode share for anything.

  • ohnonononono

    “Crowding and ridership” cause 40% of subway delays, but are all crowding delays during the 5-10% of the day when the MTA is running the tightest rush hour headways they can with the current trains and signal systems? (And let’s not get into the fact that headways were better in the past…)

    Or are a decent amount of them actually during the other 90-95% of each day when the MTA decides to run less service based on its own passenger loading guidelines? As you say, the amount of service provided isn’t adequate, “effectively guaranteeing more crowding and more delays.” The MTA doesn’t seem to want more ridership. Riders get the hint.

  • Joe R.

    I wouldn’t expect subway ridership to correlate with growth in jobs. If most of the job growth was in the outer boroughs then many of those jobs are simply not amenable to subway commutes. This probably underscores the need to expand the system. Unfortunately, I’ll bet quite a few of those new jobs result in more car commutes. Traffic in places like eastern Queens seems to have grown way out of proportion to the population. Now I see heavy traffic on the LIE when I’m out riding at 5 AM.

  • Alexander Vucelic

    key word ‘region’


  • ohnonononono

    Of course some people are making the same wage they were in 1989, but the typical person’s income has increased over the past 27 years since we had a dollar fare (which is now almost $2 in inflation-adjusted terms).

    The typical New Yorker has to take transit to work, and the $117/month in transportation costs is an astounding value to have access to millions of jobs, on top of opportunities for culture, recreation, education, etc. It’s far lower than the cost of car ownership to get to all those things.

  • ahwr

    Limit to Paris and walk mode share for all trips is 53% (by resident, similar to the numbers for NYC I gave above). Residents of Cœur d’agglomération (as close a comparison to NYC outside Manhattan as I can give you) walk mode share is ~43%. Still no idea where you get the uncited 60% figure from.

  • Bobberooni

    Don’t forget that NYC has three commuter rail systems that take people to jobs, many of them within walking distance of Penn Station / PA / GCT. I work in NYC; I don’t drive there OR use the subway.

  • Bobberooni

    Access to millions of jobs doesn’t mean much when YOUR job pays minimum wage.

  • mfs

    The graph shows otherwise and is very clearly correlated with job growth. It’s a very very strong correlation standard in many areas of economics. It’s not just commuting- people have more $ to spend on discretionary trips when jobs are up.

  • mfs

    @komanoff – can you post link to stats?

  • Joe R.

    Yes, of course the subway fare still represents a relative bargain compared to the alternatives. At the same time though it’s probably close to an hour’s take home pay if you’re making the $3.75 an hour minimum like some people I know. While on that subject, back when I was making $7 an hour in 1989, living in a double fare zone, I almost always walked 2.5 miles to/from the subway station. Paying double fare at the time would have been $20 a week, or close to 10% of my take-home pay. I just couldn’t afford double fares.

    Also note NYC has a significant number of cash workers making less than the minimum wage. My friend who owns a taxi meter shop knows of some of his competitors paying people $2 an hour off the books. Yes, it’s illegal but that doesn’t negate the fact a round-trip subway fare can represent nearly 3 hours wages for a segment of the population.

  • Joe R.

    That’s assuming the jobs are in Manhattan. My comment above was based on many of these new jobs being in parts of the outer boroughs not convenient to any form of mass transit.

  • Bobberooni

    We don’t know where the jobs are. In the absence of knowledge, there’s been a lot of speculation about why jobs have increased more than ridership. I added something that no one else had (yet) speculated upon. My guess is that most of the speculations are probably true.

  • Joe R.

    You’re probably right. I vaguely recall reading somewhere that most of the job growth in NYC in the last decade has been in the outer boroughs. Of course, parts of the outer boroughs are still transit friendly. Without knowing exactly where these jobs are, it’s anyone’s guess why jobs have increased more than ridership.

  • kevd

    4) Ain’t no more room.

  • Komanoff

    Which ones, the annual trips whose differences are shown in the graph? Sure. I can email you the sheet this evening, or create and post a Web link in the a.m. Tell me which. Or email me (kea at igc dot org) if it’s something else you want.

  • mfs

    I approximated @komanoff’s data. the below scatter plot shows year-to-year change in thousands of jobs (X) by year-to-year change in millions of subway riders. Komanoff is saying the red point (2015) has very high job growth but very low ridership gain, and indeed, it is an extreme outlier. Currently with that point, the r^2 is 0.45, without it, it is 0.55, which is a big difference.

    If anyone has a time series of monthly subway ridership, would love to do a month-to-month graph of this. As @FerdinandCesarano:disqus pointed out, a year is a big chunk of time to identify these kinds of trends with. I think the lag-lead relationship has a ton to do with the time resolution used here.

  • Joe R.

    I tend to agree we need more fine-grained data here. It may well be the years which saw a drop-off in ridership had a few months of really inclement weather which bought down the average, even if in most months the trend of increasing ridership continued.

    Another trend here might be telecommuting. Telecommuting still isn’t widely offered as an option but it probably is more available now than it was a decade ago. If so, an increase in jobs may not necessarily correlate with more subway ridership.

    In the end all we can probably do is speculate on the causes. There are quite a few variables here, each one of which can easily account for those year-to-year differences.

  • Komanoff

    My post attempts to explain/interpret the *change* in subway ridership from 2014 to 2015, specifically an ostensible shortfall of around 30 million trips. Citibike logged 1.34 million more trips in 2015 than 2014. If half of those trips would otherwise have been made on the subway, then the increase in Citibike use was responsible for less than 700,000 “lost” subway trips, or just around 2 percent of the loss.

  • Alexander Vucelic

    cititbike to private bike ratio is approx. 1:4

    if 1/2 trips substitute for subway what are other 1/2 of trips substitue for ?

  • ahwr

    How many would have been made on privately owned bikes?

  • Andrew

    Service is already maxxed out on most lines on weekends, as a practical matter, given the typical sort of construction work that goes on on weekends (locals and expresses sharing tracks with flaggers slowing trains past the work). The only thing worse than a crowded train is a crowded train mired in congestion for miles upon miles upon miles. Weekend trains have been overcrowded for years, as the MTA has readily acknowledged. The lines that don’t share corridors with other lines – the 7 and L – have seen multiple weekend service increases in recent years.

    This is not what “decides to run less service based on its own passenger loading guidelines” looks like. With one exception (bringing its loads to a whopping 81% of guideline), each of these 12 line items is an increase. When loads increase, loading guidelines are generally used to increase, not decrease, service.

    Guidelines, incidentally, don’t call for anywhere near crush loading – they call for 3 square feet per standee during rush hours and 125% of a seated load at other times. And, despite what a lot of people seem to think, a train with people packed in tight by the doors is not overcrowded if there’s plenty of empty room away from the doors – say “excuse me” and walk past them (no, you are not being rude for pushing – they are being rude for blocking the doors).

  • Alexander Vucelic

    good question – we do know that cycling and walking trips are growing year over year signifucantly. We do not know how many of these newly actuve people used to use subway or drive.

    I do know that my subway and driving use has plummeted over the last decade as cycling infrastructure was established:

    30% of my trips within Manhattan used to be in my private car. 50% by subway. 10% by cab. 20% by bike.

    Today the ratios are roughly

    <3% trips by private car or cab or subway or bus within Manhattan
    1% trips by walking
    96% trips by cycling

    It's possible that the Critcal Mass effects are growing are encouraging ever greater mode shift towards active transportation .

    Once people start cycling in the city they usually continue to use & expand cycling my in their mix of modes.

    Alone went from a hellish motor sewer to civilized city in the course of a decade. Nearly 50% of roadway traffic in London consists of cyclists. The infrastructure does not exist here, but it's getting better every year.

  • Komanoff

    Walk, taxi, bus, private car, personal bike, or trip not taken.

  • Guest

    I’d be curious to see peak/off-peak ridership levels, as well as splits in fare medium.

    There is a distinct possibility that as the unlimited 30-day card has offered less benefit, more people have opted to use pay-per-ride cards instead. Then when making marginal decisions about discretionary trips, they would end up reducing their total usage.

  • Alexander Vucelic

    I realize there are no good numbers on the cycling-other substitue process, but what do you think are the likely splits ?

    I’d guess bike trips are a substitue for::

    15% Private Car or VFH
    15% Walk
    20% Bus
    50% Subway – Commuter Rail

    but this is pure guesstimate, welcome to other thoughts on the matter.

  • Komanoff

    There’s a fair amount of data on both matters, in my “BTA” spreadsheet.

    Peak/off-peak hour-by-hour splits (for trips to or through the CBD, which is 80% of trips) are in the “Subways” tab. Fare medium splits are in the “Transit” tab.

    To access the spreadsheet, Google “BTA 1.1”. The first item directs you to a page (“Kheel Plan”) from which you can download it.

    Comments welcome.

  • datbeezy

    Something tells me… that we might be able to come up with better numbers by looking at entry point data. No 2015 data (yet) [], but my suspicion is that stations with the biggest lack of gains will be clustered.

  • thomas040



Subway Ridership Hits 65-Year High. Does Cuomo Care?

Subway ridership hit a 65-year high in 2014, serving 1.75 billion trips last year, the most since the New York City Transit Authority was formed in 1953. That’s an increase of 2.6 percent over 2013 and 12 percent since 2007, according to the MTA. The subway now serves 5.6 million passenger trips on an average weekday, and 6 million on an […]

Cuomo’s $320 Million Transit Cut Could Cost NYC Dearly

Albany’s latest raid of transit funds could hit New York City particularly hard. To help pay for his upper-middle-class tax cut, Governor Andrew Cuomo and the state legislature are stripping an estimated $320 million a year in revenues from the MTA payroll tax. Although the legislation is said to contain a pledge to find equivalent […]