Wednesday, March 18, 2015

Mapping Gov on the T

There's a great grassroots group which is pushing for Massachusetts state legislators, who live within an hour (*) of the State House, to take the T to work on March 19, called Gov on the T. They have a list of members of the General Court, and have asked for commitments as to whether the will take the T to work. I decided that this begged to be mapped, so I did.

First, a couple of words (and the asterisk above). I have no idea how they calculated an hour, but it doesn't make much sense, and there's a fragmented area that is represented by the officials they've asked. For instance, the representative from Billerica is not listed, but lawmakers from Lowell are. The Senator from Gloucester is on the list, but not the representative. The representative from Bridgewater is on the list, but the one from Hingham is not. I think it would make much more sense to apply this test to any Senator or Representative who serves constituents in a town within the MBTA's service district, which I've outlined on my maps.

(GovOnTheT, if you are reading, this, I am happy to select these districts and give you a list.)

Anyway, here is a map of House members who have pledged to take the T:

And here is a map of the Senate:

I think this is great, and partially because it is very grassroots: it's not being pushed by a lobbying group, but by a couple of people who are pissed off that their legislators all drive because they have free parking on Beacon Hill. I hope that Gov on the T becomes a regular happening, and that legislators are held accountable. Those who don't take the T—and vote accordingly—ought to get primary challenges, especially in districts where many of their constituents depend on transit. Take a look at some of the districts where representatives aren't taking the T. There's Lynn, Revere, Winthrop and the South Shore. And districts where the representatives are MIA, for instance: much of Cambridge's Senate representation and house members in Newton, Brookline and most of Boston. Let's hold our representation accountable.

Friday, March 6, 2015

ONE NIGHT ONLY! The T will provide 24 hour service this Saturday

It's kind of gimmicky, but this Saturday, the good ol' MBTA will be pretty darned close providing 24 hour service on some routes. The reason? Well, to start, late night schedules, but mostly because it's the beginning of daylight savings time. Here's the alert from the T:
Saturday: Despite the start of Daylight Savings Time, the number of available Late Night trips will remain the same Sunday morning (March 8). Last trains will depart from downtown at 3:30 a.m. with outer connections following later.
At 2 a.m. on Saturday, the clocks will magically advance to 3 a.m. The MBTA services running will not magically disappear in to the ether, but the T will basically assume that the time change won't take place until the end of service. So some of the latest-running trains and buses, which normally don't finish their runs until about 3:00, will actually not reach their terminals until 4:00. Notably, the last 28 bus will reach Mattapan at 4:05 a.m., the last Mattapan trolley will arrive around the same time, and several other lines will operate until about then.

Most T service doesn't begin until 6 a.m. on Sunday mornings, but a couple of lines, notably the aforementioned 28, operate early airport service. (There should be a discussion that better service should be provided to the airport, which has expensive and limited parking, many low-wage jobs, and many early shifts, and which is completely inaccessible from most of the city by walking or bicycling, because ocean.) So service at Mattapan this Saturday will include:
3:00: inbound 28 bus departs *
3:15: outbound 28 bus arrives
3:20: inbound 28 bus departs ‡
3:40: outbound 28 bus arrives
3:45: inbound 28 bus departs *
3:59: inbound 28 bus departs †
4:05: outbound 28 bus arrives
4:45: inbound 28 bus departs †
* Saturday late night service   † Sunday AM service   ‡ Trip scheduled on both Saturday late night service and sunday AM service (!)
Now, it's worth noting that there is a service gap of nearly three hours in outbound service: the first outbound bus on Sunday isn't scheduled to arrive in Mattapan until 6:40. But what I find most intriguing is the fact that for nearly an hour, the Saturday and Sunday service actually overlaps (on a normal weekend, the last outbound arrival comes in just 15 minutes before the first inbound departure). Most interesting: the 3:20 a.m. inbound trip can be found on both the Saturday schedule (the 2:20 inbound trip bumped an hour) and the Sunday schedule (the regular 3:20 departure). Will the T run two buses inbound from Mattapan simultaneously, one on a Saturday schedule and one on a Sunday? It's almost worth venturing down to Mattapan to see.

These are strange times we live in indeed.

Tuesday, March 3, 2015

Bus efficiency: Passengers per Bus

I've been intrigued by a few statements and metrics regarding buses recently. The first is that Beverly Scott recently said (on the radio, I think) that the T really should have many more buses than it does; she'd expect an agency the size of the T to be running 1500 buses, instead it has only about 1000. The second is that the T can not add any buses to the fleet, not only because of the cost of acquiring and operating new buses, but because there is physically nowhere to store and service them: the current bus yards are maxed out. And with proposals floating around for post-Sandy-New-York-style bus bridges, there would need to be a lot more buses in the fleet for that to happen.

My question was how heavily used the T's buses are, especially compared to other systems. For every [fill in the route number here] bus that is jammed full of passengers, there's a late-evening 350 with just a few riders. I decided to look at the ridership and fleet sizes for large bus fleets in the US, and see how many daily passengers there are per bus in the system. (Note that most fleets carry a 20% spare ratio, so the actual number of passengers a bus will carry in a day is 20% higher.) The short answer is that the T falls in to the normal range of major systems, if even at the lower end. The longer answer is that each system is different, and comparing them is all comparing apples to oranges.

You came here for charts, right? Here's one. On the right axis are US bus systems with more than 200,000 riders per day. On the bottom axis, the number of riders per bus. I'll get to the colors below.

So note that the systems in San Francisco and Chicago perform way ahead of anywhere else. The reason there is twofold. One is that in both systems, buses are the backbone of the system: they carry more passengers than the cities' rail systems, so that many of the busiest corridors are served by buses. In fact, of the top tier of transit cities (there's a big drop from Seattle, with 390,000 riders, to Baltimore, with 250,000), Boston, New York and DC are the only cities with more rail riders than bus riders. The MBTA's bus system really operates as a transit feeder system (and always has); the only "Key" bus route in Boston which serves Downtown Boston—other than the Silver Line—is the 111. Most of the busiest bus routes—the 1, 28, 39, 66 and so forth—are either crosstown routes or feed a transit line.

The other difference on the above list is that in several cases, the bus systems primarily—or only—serve the city, with a separate system (or systems) serving outlying areas. For the top-eight systems (in red), only the MBTA, SEPTA, WMATA (DC) and King County Metro (Seattle) serve the whole area, the rest are augmented by suburban systems (Pace in Chicago, several agencies in LA, New York and the Bay Area). For Chicago and San Francisco, I added in other agencies (Pace in Chicago, AC Transit, Golden Gate Transit and SamTrans in San Francisco, since all serve the City there; shown in orange) and it brings their passengers per bus rate in line with other systems.

In fact, it's interesting that once this adjustment has been made, there are no high outliers: the top systems fall between 375 and 460 riders per bus. There are outliers on the lower end. WMATA is lower by quite a bit, and Seattle lower still; it's possible the fleet size data there I am using isn't perfect, or that they have different utilization rates. In DC, there are poor transfers between buses and trains, and rail service carries the bulk of passengers. I'm not sure about Seattle, but it has a very small rail system relative to any other system on the list.

In any case, it's all apples to oranges because Boston is relatively unique among these systems. The buses are mainly set up as rail transit feeders (this is the case really nowhere else, at least to the extent that BERy was built on subway-surface stations), the system is region-wide, so lower-ridership suburban routes skew the numbers and buses do not carry the majority of riders. If this was a Venn Diagram of cities with more rail than bus passengers and a single region-wide bus and rail provider, only the T and DC would be in the middle. And the T does far better at bus utilization, it seems, than DC, and while Boston's system was mostly built to integrate surface and subway lines, DC's Metro was built long after the bus lines had already been established (mostly as streetcar lines, of course). So with the nearest we can get to an apples-to-apples comparison (apples-to-pears, perhaps?), Boston is far more efficient.

So, the T needs more buses. Except for San Francisco, the T, at ±1000 buses, has by far the smallest bus fleet of these top-8 cities (and San Francisco—with 800 buses—only has a 50 square mile service area; about the size of Boston proper; AC Transit alone has another 550). Cities outside the top-8 don't have the same sort of capacity constraints the T (and other major bus systems) see on a daily basis. In Minneapolis, for example, the busiest bus route runs every 6 minutes. In Boston, several routes—the aforementioned 111, the not-even-key 7—run every 3 to 4 minutes at rush hour, and they're packed. 

Having more buses would allow more operational flexibility: a route like the 70, which could see many of its problems solved with a couple of extra buses, could see such service without worrying about where to put them. But that's easier said than done: finding land for a new or expanded bus depot isn't an easy proposition, especially when existing land, in many cases near transit nodes, has high value. Still, ordering new buses is faster than new rail cars (one year lead time instead of half a decade) and having enough vehicles would have a positive impact not only during irregular operations, but on routes which are overcrowded right now. Buses can be procured off the shelf for $500,000 each. A $50 million capital investment for 100 buses—plus the cost of space to service new buses, of course—could have a quick but lasting impact.

Monday, February 23, 2015

Go to public meetings

A chart that looks conspicuously like one which has
appeared on this page.
I went to a public meeting tonight regarding the 70 bus. I go to a decent number of public meetings, and this one piqued my interest because of my interest (and use) of the 70. So I trudged out in the 5˚ temperatures to Watertown to talk bus schedules.

So did a couple dozen other people. But, of course, they didn't want to talk about the nitty gritty of bus schedules. They wanted to air their completely unrelated concerns. Melissa Dullea, who manages to not roll her eyes back when people yell at her about completely unrelated topics, was great, and the conversation was steered slightly in the direction that it needed to go. She also gave me a shout-out and used a modified version of my 70 bus post to illustrate the headway issue on the route.

But we also got to discuss such important items as:

  • Whether the T was going to eliminate stops at senior housing
  • How the straps on the buses aren't low enough and that's why there isn't enough capacity
  • That people don't move to the back of the bus and could drivers please make announcements more often (why people don't take initiative and ask themselves is beyond me)
  • It would be great if there was a device that would tell you when the next bus was coming that didn't require a smart phone. We have those. They're called smart phones. (The best exchange was when one guy said "it would be a big seller" and someone else said "no, it wouldn't.)
  • That those straps sometimes are missing completely and what can you do about it. (Why not tell the T the bus number.)
Just another day riding the 70. Or, as I like to call it, the 140.
What I'm trying to point out here is that if you don't go to a public meeting, these people will. It's not that their concerns aren't valid, but they're noisy, and completely uneducated. It's kind of like NIMBYism: they care only about what directly affects them, with no concern for the greater good. And then when there's actually an issue: a poorly-timed bus route, like the 70, they service planners will think that no one actually cares, because all they hear about are bus strap heights. I'm serious. Go to public meetings. 

There was a great presentation later about an app ( which will allow people to report bus issues to a central database. Right now it is Watertown-specific, but I think it would be portable to the whole system. I love user-generated feedback like this. Now if there was only a device which would let you find out when your bus was coming that wasn't a smart phone.

I left a few minutes early to catch the 70. And of course, two buses came at the same time.

Wednesday, February 18, 2015

Banned in Boston: a graduated income tax.

WBUR ran a poll asking about the MBTA. They asked how people rated the MBTA (78% fair or poor; or what it's worth, I'd rate it "fair") and how often it was a reliable means to get around (61% "most of the time" or "always"). Nearly everyone (85%) thinks that fixing the T should be a priority, and most blame the outdated system rather than management (66%, although Republicans are more likely to blame management).

They then asked people if they would support additional taxes or fees to help fund the T. 48% said they would. 48% said they would not. (Those who supported it split between somewhat and strongly support, although opposition is skewed towards strongly oppose.) The party crosstabs are somewhat interesting: even 30% of Republicans would support new taxes or fees, more than half of those strongly.

Most of the poll's crosstabs don't show anything particularly interesting, and I think the pollsters made a mistake by not asking people how they typically commuted, as that would have been a very interesting crosstab (I talked to the pollster and it may be forthcoming). However, what stands out is how the spread in responses varies. I expected it to vary by geography: people in Boston or "inner core" areas would support more taxes, or be more likely to blame an old system rather than wasteful management. But that's not the case. Instead, the responses to these questions varies by income and education, which I assume are well-correlated.

For instance:

Q: "Which of the following do you think is most responsible for the issues this winter?"
A1: Old system and Maintenance
A2: Poor management

     Overall  under 25k  25-75k  75-150k   150k+
A1:     66        53      65       72       74
A2:     17        22      17       16       10

Wealthy voters are far more likely to blame the old system than poor management. For low-income respondents, they're only a bit more than twice as likely to blame the system than the people. For higher income respondents, they're seven times as likely. There are similar trends when you ask about funding shortfalls:

Q: What is responsible for the system's financial shortfall?
A1: Not enough legislative funding
A2: Waste and mismanagement
A3: Prioritizing expansion over maintenance

     Overall  under 25k  25-75k  75-150k   150k+
A1:      30       19       25       35      39
A2:      36       41       36       39      32
A3:      23       31       29       22      18

Here, a high-income respondent is twice as likely to blame the legislature than a low-income one.

Q: Would you support paying more taxes/fees for better transit?
A1: Strongly/somewhat support?
A2: Strongly/somewhat oppose?

     Overall  under 25k  25-75k  75-150k   150k+
A1:     48        40       44       50      63
A2:     48        51       55       48      35

Here's where I think it's interesting. Wealthier respondents are far more likely to support more taxation. By a nearly two-to-one margin, people making more than $150,000 a year support more taxes, while people making less than half that much are against it. Herein lies the problem: how do you tax the people who make more—and are willing to pay more—without raising taxes on people struggling to get by.

The obvious answer is a progressive income tax. Massachusetts is one of few states without one. The problem with changing the tax structure is that progressive taxation is explicitly forbidden by the state constitution. Oh.

(Update: here's a good primer on the state income tax with more up-to-date numbers.)

Right now, the standard deduction of $8800 for a family makes the tax structure is slightly progressive. I've calculated the tax paid under the current system and a proposed progressive system for various incomes (5.15% after the standard deduction). On the left are the median incomes for each quintile of earners in Massachusetts (from 2006, but the numbers haven't changed too much since then). On the right are what would happen if you raised the standard deduction to 10,000 and had the following graduated rates: 4% to $25000, 6% to $75000, 8% to $150,000 and 10% on income above $150,000.

Income       Tax       Rate     New Tax    New Rate
$20,000      $577      2.8%     $400       2.0%
$48,000      $2019     4.2%     $1580      3.3%
$75,000      $3409     4.5%     $3200      4.3%
$105,000     $4954     4.7%     $5600      5.3%
$175,000     $8559     4.9%     $11700     6.7%
Total/Avg*   $19518    4.2%*    $22480     4.3%*

(* Average of five rates here, not the average rate paid for the total)

Taxpayers in the lowest two quintiles of earners would see dramatic decreases in their taxes: $200 to $500 (25-35%), which, for lower earners, is likely to be spent, boosting the economy. The middle quintile would also see a small tax break: someone earning $75,000 would get a $200 tax break (5% decrease). Upper earners would see their rates increase, but the change would not be more than $1000 until it was well in to six figures (a million dollar earner would see their tax rate increase from 5.1 to 9.4% and pay an extra $43,000, but I think they'll still get by).

For the Commonwealth, it would mean new revenue. Based on just these numbers, there would be an increase of 15% from income taxes. But the math is much more complicated than that involving calculus I'm not about to do here since the long tail of high income earners would pay more while the decreases at the lower end of the scale are more minimal, so the revenue increase would likely be a bit higher. So, most Massachusetts residents would get a tax break, and the Commonwealth would have more money to improve services for everyone. And according to the survey, higher income residents would be willing to pay more. It's a win-win-win, but it can't be implemented without amending the constitution.

There are two ways to get this on the ballot. One involves a majority of state legislators and is probably a non-starter: Beacon Hill types are notably averse to doing, well, anything. The other way is for 3% of Massachusetts voters would need to sign an initiative petition, and then 25% of two consecutive legislatures would have to approve it. 25% of two consecutive legislatures might actually have a chance. Then it would go to the voters.

The last attempt to make this change in 1994 (a Republican wave year) failed by a wide margin, but an attempt to end the income tax all together in 2008 failed by a wider margin. The political hurdles would be large, but it is clear that the state needs more money to invest, and it is clear that higher earners are willing to pay more taxes. Convincing everyone to vote for this—even if it would be a tax cut for most—would be the hardest part.

And to the naysayers who crow that it will drive away jobs, look at Minnesota. There, the governor raised the top scale of the income tax and turned a deficit in to a budget surplus, which is being reinvested in schools and infrastructure and targeted tax breaks for the working poor. Too bad that it's a well-educated, progressive state that happens to get a lot of snow: does that sound familiar?

Tuesday, February 17, 2015

Tallying snow and cold in Boston

I'll update this scorecard for the next few weeks.
Follow @ofsevit on Twitter for the latest version
A few weeks ago, I wrote about how unprecedented the weather has been in Boston, and how it affects the transportation system. How unprecedented is this February? It is the confluence of two 500-year events: temperature and precipitation have both experienced anomalies at least 3 standard deviations from the mean. It is without parallel.

We covered snowfall in the last post, here we'll focus more on the cold temperatures.
There has been one month, in the recorded climatological history of Boston, where the average temperature was below 20 degrees. 1934. It was unprecedented cold: the all-time record-low was set in Boston at -18 (a temperature which, with the observation station being moved to Logan Airport in 1936, will likely never be matched; the airport has only recorded three days below -10, the most recent in 1957). The average temperature in February 1934 was 17.5˚. The next highest months were Januarys in the late 1800s when the temperature averaged 20.1 (records date to 1872). Winter temperature records are very nicely normally distributed (69/96/99.7; other months are similarly distributed but there is much less nominal variability in the summer), and there has only once been a a month more than 3 standard deviations from the mean out of more than 400 winter months. 1934.

It is quite likely that eight decades later, 2015 will be the second. The dreaded polar vortex last year gave us departures from the mean of 2 to 3 degrees. (To be fair, we were on the edge of the vortex last year, and the Upper Midwest had similarly anomalous weather.) This year, we're likely to run more than 10 degrees below normal. The average high for mid-February is 40˚. We haven't seen that temperature in nearly a month, one of the longest stretches on record.

Thus far, temperatures in Boston have averaged 18.2 degrees, and if the forecast for the next week holds up, the average will actually fall below the 1934 record. With five days beyond then in the month, even if we revert to seasonal climatology normals (unlikely, given the current pattern and modeling), we’d set the second coldest month on record. If the weather stays cold, as is advertised, we’ll see the second winter month more than three standard deviations from the mean. And we might break the all time coldest monthly record. Which, given climate change (mean temperatures have risen 3-4 degrees since 1872), is particularly impressive.

Apparently it has snowed quite a bit as well. And, yes, the seven feet in three weeks has never happened before (and may well never happen again), and the past month has already outpaced any winter season in Chicago, New York or DC (and is one storm away from topping Minneapolis). But without the cold temperatures, it wouldn’t have had the same effect: some snow would have melted even with a few days over 40˚. In 1978, days in the 40s and 50s helped reduce the snowpack to 4" before The Blizzard. In 2013, a week in the 40s reduced a major blizzard to a few inches of crust. This year, we haven't seen 40˚ since MLK Day. The snow combined with the temperatures is really a double-whammy: two once-in-500-year events, at the same time.
From Sam Lillo @splillo

It's something no one could have planned for, and something we'll never see again. And I do need to update one of the charts in that first post: the 28 day snowfall. The current chart is here. But it may increase.

Update: According to Sam Lillo, who has some great posts like this one, Boston has had more snow in the past month that Buffalo ever has. Will try to confirm (confirmed).

Friday, February 6, 2015

Commuter Rail Ridership and Fares

A Twitterer recently found a 1972 plan for transit-level service along many of Boston’s rail corridors and was floored that ridership on what we now know as Commuter Rail at that time was only 16,000 per day (it was 80,000 in the '40s and now hovers around 70,000). It has indeed grown, especially the South Side lines: in 1972, only 600 people rode the Worcester Line daily (the then-speedy Mass Pike having recently opened); many single Worcester Line trains now carry that many; the line has grown more than 15-fold over the past 40 years.

But the time of impressive growth is even more impressive: in 1981, ridership on Commuter Rail still hovered around 17,000. And then it began to grow. By 1990, it had more than doubled. And during the 1990s, it more than doubled again, so that by 2000 there were more than four times as many riders as there had been two decades before. Some of this is attributable to extensions after service cuts in the ‘70s, and new service on the Old Colony Lines. But a lot is due to the revitalization of downtown Boston following the growth of the 128 corridor, worsening traffic, and higher parking costs.

For 22 years, from 1981 to 2003, Commuter Rail traffic grew every year but one: 1991, after the collapse of the “Massachusetts Miracle.” But it took off again thereafter, peaking at 74,000 daily riders in 2003.

It hasn’t been that high since.

At first, flat ridership could be blamed on the early-2000s recession. But in the past 15 years, ridership has stagnated. If ridership growth had continued, linearly, at the 1981-2003 rate, it would be poised to cross the 100,000 threshold this year. Instead, the numbers of riders has barely budged, fluctuating up and down as the economy, traffic, and the price of gas has ebbed and flowed, none of them seeming to dramatically affect ridership.

Except for fares. Take a look at the chart. It certainly seems that, once fares started to rise dramatically, ridership flattened out. In the last fifteen years, commuter rail fares have gone up 250%, while they didn't rise that much in the 20 previous years (despite higher inflation). Subway fares have risen as well, but the nominal amount the fares have risen is very different.

In the '80s and '90s, a ride on the T cost 60 or 85 cents (that's a token, by the way) while Commuter Rail fares ranged from $1.75 to $3, rising to $2.25 to $4. The ratios were similar to today (the highest Commuter Rail fare about five times a subway fare), but the difference only $3. Now? The difference between a tap of a Charlie Card and a punch of a ticket is $8, which is a much greater difference.

There are two salient bits here. First, subway ridership, despite the same relative rise in fares, has seen dramatic increase in passenger counts in recent years, despite the increases in fares. At issue here is the fact that Commuter Rail and urban rail are different populations: Commuter Rail passengers have more options. Most own cars. If the cost of driving and parking is not much more than commuter rail—and parking and driving costs haven't more than doubled in the past 10 years—they are more likely to abandon the rails and head for the highways. If the MBTA provided excellent rail service, with fast speeds and reliability, this would be less of an push factor. But with old equipment and slow track, it is.

The second piece is that outside of peak travel times, the train generally can't compete with vehicles on travel time, and on weekends, on ease or cost of parking. And while many trains at rush hour are near capacity, there is plenty of capacity on off-peak and weekend trains. It's possible that if the T offered off-peak savings—say, $2 off all fares beyond zone 1A, or half off, or something—they could drive enough additional ridership to cover the lost fare revenue, all the while taking cars off the road, which is good for everyone. As would having sensible, clockface midday schedules. And it might even help ridership trends.

It's rather obvious that Commuter Rail ridership is more elastic than subway ridership: when fares go up, ridership might not go down, but previous trends level off. Since urban riders would cry foul if local fares rose faster than Commuter Rail fares (the subsidy per ride is higher for Commuter Rail riders, although the subsidy per mile is about even), increasing Commuter Rail ridership would require better service. Given the recent performance of the Commuter Rail, this may be a tall order. But it should be a goal.