Subway “Summer of Hell” Really Started Years Ago, Data Shows

The four-day span from Friday, April 21, through Monday, April 24, may have been the tipping point for the New York City subways. A midtown power outage on the 21st severely delayed half the subway’s lines, creating a nightmare commute on the same day that a switching problem left NJ Transit riders stuck on trains for more than an hour. Three days later, three separate incidents during Monday’s morning rush hour resulted in another round of misery.

Since then, it seems, the subway has yet to recover any semblance of normalcy. The following month brought three similarly hellish days. On May 7, service on several lines running through Brooklyn’s Dekalb Avenue stop was severely disrupted by a Con Edison power issue. A few days later, almost a dozen lines experienced rush hour issues. The following week, signal problems and “fallen debris” disrupted some half-dozen lines. June greeted New York with the infamously trapped F train riders, whose harrowing ordeal caused every New Yorker to let out a collective “Hell no.” By the end of the month — after an A train in Harlem hit an unsecured piece of track and derailed, injuring 34 people — Governor Andrew Cuomo had declared a “State of Emergency” for the MTA. But that was just the icing on the severely delayed cake, perhaps due to cake traffic ahead of us.

To most commuters, the disasters this spring felt like a new era of MTA mismanagement. But this wasn’t for a lack of alarm bells. In February, the New York Times parsed the MTA’s train trip data and found that the average number of subway delays per month had more than doubled since 2012. There was already a palpable sense matters were getting worse — one rider told the Times, “The last few months have been maddening” — but the difficulties remained nuisances, not yet crises. When the Voice reported two years ago that the MTA was facing “three or four decades’ worth of disrepair and disinvestment,” the worry was still over headaches like frequent weekend service changes to the L train, not unplanned shutdowns on a near-daily basis.

“I do believe the system has gotten worse this year,” says Rich Barone, vice president of the Regional Plan Association. “I wouldn’t disagree with that.”

Even the MTA admits service has declined in recent months. But how much worse it’s gotten — and why it’s happened so quickly — are not as easy questions to answer. The lack of clarity speaks to a fundamental issue with the MTA’s management: It isn’t measuring the system’s own performance in a meaningful way.

The MTA uses two metrics for punctuality, as explained by Alon Levy in Vox last month: on-time performance (OTP) and wait assessment (WA). OTP is the percentage of trains that make their final stops within five minutes of the MTA’s schedule, which has not been updated to reflect speed limit and other operational changes over the past few decades; WA is the time between trains. For every delay, the MTA must record a reason, such as overcrowding (which, Levy notes, is a catchall used by train operators when the true cause isn’t identified), signal failure, or equipment malfunction.

Neither of these metrics tells us very much, in part because they’re binary: The train is either on time or it’s not. To be sure, a train stuck in the tunnel for 45 minutes backing up the system for hours during rush hour and creating dangerous and sweaty crowds on platforms leaves a footprint — one mechanical failure on a moving platform at Union Square in 2015 resulted in 625 separate train delays — but even those raw numbers downplay the actual effect it has on the system and the misery it creates among New York’s millions of straphangers. These statistics count trains, not passengers, meaning each trip counts equally, whether the train is running at 2 a.m. on a Tuesday or rush hour on a Friday.

A more accurate way to measure the severity of delays would be to follow the lead of London’s transit agency and use Lost Customer Hours (LCH), which illuminate the magnitude of the most disruptive delays. For example, a derailment on London’s Central line on March 10 caused a whopping 136,663 LCHs; only one other incident that month caused more than 15,000 LCHs. We can measure that Central line derailment against other delays to determine its true magnitude. But not so with the MTA’s problems.

Take that disastrous 625-delay day in 2015. Even 625 trains are a fraction of the thousands that ran that day. In the end, it resulted in only a tiny blip on the MTA’s monthly performance metrics. According to a New York State comptroller’s office report published that year, the subway logged 498,889 delayed trains from March 2013 to March 2014. Adding 625 additional delays to an average month would bump up the overall number of delays by a mere 1.5 percent. (Of course, it is a much, much smaller percentage in terms of on-time performance, which measures delays against all trips.) In all, it’s not much more than a rounding error, because they’re measuring the wrong things. Trains running on time is not what concerns people as much as hours lost or customers inconvenienced.

“The fact that there aren’t a lot of metrics that can help the public determine the source or severity of delays is a real problem,” Masha Burina of the Riders Alliance tells the Voice. But for the MTA, measuring LCHs or a similar metric would demolish several convenient fictions. For one, the MTA maintains that the leading cause of delays is overcrowding, but if the public could see the outsize effect mechanical issues have on major delays, the MTA would no longer be able to cast itself as a victim of its own success.

And once we the public know the scale of disruptions — how much worse, say, a midtown power outage is than a run-of-the-mill Tuesday overcrowding — we can further diagnose the service’s deterioration as it affects millions of commuters.

(The MTA did not reply by press time to queries about why the authority doesn’t track lost customer hours.)

Which brings us back to the original question: How much worse is the subway getting, and by how much, how quickly? Instead of providing answers, we must make educated guesses.

Absent objective data, the best we can do is analyze media reports of major delays. This is far from a perfect method — subway panic begets more coverage of subway panic — but even the public perception that the system has reached an extreme not previously experienced can be a useful insight. And the late April/early May demarcation point coincides with Burina’s anecdotal evidence gleaned from speaking to Riders Alliance community members. “That was when it went from a nuisance and grievances to a real crisis point,” she says.

So, why April? What was different about April that made it the point of no return? The answer, unfortunately, is an unsatisfying one.

“This problem is years in the making,” says Barone, who thinks the MTA’s backlog finally caught up to them. “It might have reached a climax, a high point — or a low point — a couple months ago, but the reality is, this is years of backlog, years of underinvestment, increased ridership, [and] pressures on the system.” As in, the same answers we’ve been hearing for years.

A quest to find a reason for the timing will prove futile, Barone cautioned: “I don’t think you can point to one single thing or action that made it sort of crumble.” As it turns out, the warning bells were right. But by the time anybody noticed, it was too late.