Public Transit Unions
In this post, I would like to explain an observation: public transit workers are unionized even in countries where few workers belong to a union, such as the US and the UK. Attempts by management to win concessions from the unions have often led to prolonged strikes, for example the 108-day strike by SEPTA workers in 1983. This is not just the public sector: the private railroads in the US are unionized, and in 1964 the Florida East Coast Railway had a year-long strike. Nor is this because the public operators and private railroads have a unionized legacy: many private bus shuttles used by Bay Area tech firms to get workers from San Francisco to Silicon Valley offices have unionized drivers, including Facebook and Google, even though the tech industry as a whole is predominantly non-union. Public transportation workers, especially on-board crew, are in a position of strength, and I would like to explain why.
First, let us recount how unions work. At their heart is collective bargaining: the union negotiates wages and benefits for the entire membership. An individual worker who is paid too little can quit, which is unlikely to cause the employer much concern; a union can launch a strike, thereby getting higher wages and better work conditions than each individual worker could obtain alone.
In order to be able to collectively bargain, there needs to be a collective in the first place, which means two separate things:
- Politically, there needs to be a sense of solidarity among the workers. The workers need to see each other as comrades, or as colleagues, and not as competitors. They need to feel like the union represents their interest. This is unlikely to happen in industries where there is substantial competition for promotions to managerial roles, but easier in industries with clear separation between line workers and managers. It is also unlikely to happen in industries with rapid turnover of employees.
- Economically, there needs to be a reasonable way to set a uniform wage scale. This means that the work done by the various employees in each category must be interchangeable, or close enough to it that wages can be made equal. In turn, this requires that the job not have a strong sense of merit, in which some workers are persistently much more productive than others. Productivity differences should be related to seniority, or easily worked into wage agreements (for examples, sales agents can collectively bargain for a uniform commission structure). It’s fine if there is differentiation into several job categories, but then it’s best if this differentiation is rigid, so that workers in each group do not often have to do the job of workers in other groups.
Conversely, if the union goes on strike, management has tools to fight back, including the lockout, or more commonly hiring strikebreakers. To deter management from doing so, the union can exercise peer pressure on people not to scab (condition 1 above), but it also union needs to also fulfill a third condition:
- The job must be skilled enough that it is hard for the employer to hire and train strikebreakers quickly. On a nationwide basis this makes unionization easier in a tight labor market, hence socialist support for monetary policy that prioritizes full employment over price stability. But on a per-industry basis, this condition requires some difficult pipeline that workers must go through: a degree, long training, apprenticeship, or bespoke familiarity with the project at hand.
Conditions 2 and 3 are in tension, since skilled jobs are more likely to involve workers with different levels of productivity. For example, retail workers fulfill condition 2, but are easy to replace if they go on strike and the labor market is not tight. In contrast, programmers easily fulfill condition 3, in the sense that if the entire development team on a project quits, it will be difficult for the company to find a new one and get it up to speed, but because there is a wide range of productivity levels, it makes little sense for programmers to all be paid the same. Silicon Valley’s business culture, in which developing a product fast is more important than establishing a business culture, and turnover is encouraged, also makes it hard to establish meaningful solidarity among software developers.
In the intersection between these three conditions lie a set of job descriptions in which there is a strong sense of professionalism, to meet condition 3, but a weak sense of merit, to meet condition 2. These jobs are typically static in the sense of not changing too much over the years, which makes it ideal for a person to do the same thing for their entire life, learning to adopt new technologies gradually as they are introduced but not having to change their entire skillset. Often, the industries these jobs are in are static as well, whence they are more easily done by the public sector, or by large conglomerates that have been around for generations, such as the private US railroads.
Commercial drivers – of buses, trucks, trains, planes – are a major example of people who satisfy all three conditions. One racecar driver can be better than the next by being faster, but in the commercial sector, speeds are determined by the equipment, the schedule, and the safety standards. A train driver can be better than another on the margins, by responding more quickly to an obstacle on the tracks, or paying attention to the posted schedule better, but both of these aspects depend more on external factors, such as the signal system, than on driver skill. Train schedule padding, accounting for suboptimal driver behavior such as beginning to brake too early but also for propagating delays and passengers who take too long to board or alight, consists of a few percent of total travel time: in its peer review of California High-Speed Rail, JR East proposes 3-5% (PDF-p. 10); and a Swedish study for high-speed rail mentions a 0.9-2.3% discrepancy in energy consumption based on driver behavior (PDF-p. 12) and a total schedule pad equal to about 7% of travel time (PDF-p. 24). The productivity difference, or in other words individual merit, is too small to challenge the logic of a uniform wage scale.
Instead of individual merit, commercial drivers have professionalism. A train driver is expected to fulfill certain criteria to ensure safety first and punctuality second, which requires considerable study of the route of the train, the signal systems, the dispatching codes, the correct way to respond to various unforeseen circumstances. Unlike individual car drivers, commercial drivers are not permitted to take small risks in other to go faster, and learning how to pilot a vehicle safely takes some time; all of this is true manifold if the vehicle in question is a plane, leading to arduous certification requirements. In this setting, even in an environment of absolute control by management, there’s little reason to have different payscales – at most, management can penalize workers who make mistakes.
For the same reasons that train drivers have standards of professionalism, it is easy for them to form a cultural group with internal solidarity. They have their own knowledge set and jargon, as anyone who has tried reading threads on railfan forums knows. The transportation industry changes slowly enough, in terms of both travel demand and technological progress, that most train drivers can expect to work for the same company for decades. Even bus drivers, in an industry that’s less dependent on physical plant, can expect to move between bus operators.
The other major category of public transportation workers, maintenance workers, is not as clear-cut. There is certainly a difference in merit – some people just fix things faster than others. But at the same time, the importance of safety is such that giving financial incentive to working faster can lead to shoddy work. Instead, there are extensive regulations, developed by national safety authorities as well as internally by rail operators, specifying which tasks need to be performed and at what intervals.
Unlike train drivers, maintenance workers aren’t interchangeable – they have more specific job titles. But these job titles often lend themselves to easy categorization, such as electricians and welders. Those have their own standards, and the potential for major accidents encourages uniformity of credentials and of wages, while the complexity of the machinery these workers operate ensures that they must be skilled. This leads to the same presence of professionalism without differential individual merit seen in the case of train drivers, and ensures all three conditions facilitating unionization are present.
In a healthy company, technology progresses as fast as it can given the requirements of the industry, and the union rules are reasonable. Wages and benefits are higher with collective bargaining than without almost by definition, but they cannot be massively higher – those are skilled workers, who are not easily replaceable. In environments where the rules are unreasonable – perhaps the certification requirements are more onerous than necessary, perhaps labor-saving technologies are not used, perhaps the work rules are not suitable for a modern operation – management cannot easily force the union’s hand. Even when three people do the job of one, as in the case of US commuter rail operations with conductors and assistant conductors, labor is in a position of power, and reform-minded managers cannot easily hire strikebreakers.
Pedestrian Observations 
Since this is a politicized issue, I didn’t want to put it in the main post, but we can study teachers’ unions using the same lens. Teachers clearly fulfill criteria #1 and 3 for easy unionization. Criterion #2 is murkier: teachers do have differential levels of merit, but not nearly to the same extent as programmers (or researchers), so uniform wages are not universal, stronger in the public sector than the private sector. This can be broken down into the following issues:
a) The public sector has incentives to set uniform wages and benefits, even when there are different productivity levels in play, in order to discourage corruption. Of note, the practices in public educations that ed reformers oppose, including teacher tenure, wages that depend only on seniority and credentials, and relative teacher autonomy in curriculum, are not innovations of the labor movement. In the US, teacher tenure goes back to Progressive reformers who wanted teachers to be independent of political machines. This is why there’s less uniformity, and less union penetration, in private education, and none in for-profit institutes such as test prep and language schools.
b) While there are clearly better and worse teachers, in mass-market classes, the differences aren’t very big. At UBC, I taught 100-student calculus sections, designed to weed out weaker students. There were 8 or 9 sections teaching the same curriculum to the same midterms and finals. This is not a perfect natural experiment, because the sections were at different times of day and different days of the week, clashing with different classes, so there were clusters of students from specific programs in certain sections; because of this effect, more senior professors explicitly told me not to worry if my students’ midterm averages were not as high as those of other lecturers’. With that in mind, my recollection is that the highest section averages were about 5 points higher than the overall average on a 0-100 scale, and the standard deviation for individual students was about 15, so the teacher effect looks like 1/3 of a standard deviation. Moreover, there was a seniority effect – the senior professors were persistently better than the first-time teachers. Now, again, these are weedout classes, and there is stronger teacher effect in smaller groups or in seminars, or in any situation in which a better teacher may convince a student to stick with the subject (e.g. by majoring in math if it’s an introductory class at uni, or by not dropping out if it’s a working-class high school). The exact degree to which primary and secondary teachers exhibit individual merit vs. professionalism is a point of political contention.
I’ve seen much bigger differences in teacher quality in schools where management had gone completely AWOL. My high school, for example, had several appalling teachers who could easily have been removed on malfeasance, misfeasance, or non-feasance grounds, but no principal was ever in office for more than 2 years, so the process didn’t even get started.
Nothing to do with unions really, but is shows what happens where there’s no management. The good teachers knew about the grossly incompetent teachers, but it wasn’t their job to do anything about it…
Alon,
I do want to point out the Florida East Coast railway strike lasted MUCH longer than a year(although within less than a year replacement workers had the railroad back up and running). Technically the strike was not settled until the early 1990s when the railroad returned to normalized fully union operations. The FEC strike while not well known outside of Florida was perhaps one of the nastiest and bitterest in American history right down to striking workers trying to blowup up the railroad with explosives.
https://fee.org/articles/the-little-railroad-that-could/
I will note though the many of advances both in technology and work rules instituted by the FEC in the strike years were later adopted by the other larger freight railroad and Amtrak. For example end of train devices, hot box detectors, cab signaling with ATC, paying crews by the hour not the mile etc.
I have read this piece several times now but remain puzzled as to what its main point is. Is there some argument that high wages costs (or that bogeyman, “flexible” work practices) is a major impediment to passenger rail transport? And that it is a result of those conditions that give these unions special power? Having lived for a decade in France where there are ritual rail strikes–often in peak summer–but which the country survives and the public systems (Metro, rail and TGV) thrive (arguably among the best in the world), I don’t believe it is as big an issue as many want to believe. And of course, despite strong unions Paris was still one of the first, if not the first major cities, to introduce driverless Metro trains (Meteor Line #14 1998, though first trialled in Lille years earlier 1983) and it has spread to Line 1.
The two biggest issues, at least in the Anglosphere, that make building and operating good rail transit difficult are (1) explosion in capital costs (new build or major renovation & maintenance) and (2) failure to allow PT to capture the immense value they bring to their hosts (cities, travellers, businesses etc) which would transform their economics. These have little to do with labour costs or work practices, but everything to do with a philosophy that private interests have the right to profit as much as possible from both the construction and management of such transport projects, and then all the benefits, even though publicly-funded monopolies, must flow unimpeded to private interests.
This is exemplified in an article in today’s Guardian (link below) on the worst privatization of the BR network: Southern Rail. Which happens to be one I experienced for the five years I lived in Brighton-Hove, but pre-privatization when I thought it was pretty good. Extracts:
https://www.theguardian.com/commentisfree/2016/jul/19/southern-rail-failure-crush-unions
Southern is a story of rail failure. But the real agenda is to crush the unions
This is the most farcical privatisation even by the comedic standards of British railways – and the aim is to defeat one of the last holdouts of organised labour
Aditya Chakrabortty, 19 July 2016.
So that answers any question of whether the Southern rail network is worth supporting on pure economic grounds. And whether labour has anything much to do with its current failures.
Unions can sometimes be obstructive to progress but, as we can see in most of the modern world’s best metros or commuter rail (alas few in the Anglosphere), are not a block to achieving excellent operation. Setting up a war with the workers is hardly going to be productive.
The point of the post is to argue for what you’re saying in the last paragraph of your comment: setting up a war with the workers is not going to be productive. In situations where the workers have low productivity, as in the US, or for that matters RER train drivers, it’s not really possible for management to threaten to sack workers or bust the union, because unless they’re automating operations, it will take forever to train workers. It may also take a long time to figure out better work rules without the intimate knowledge of the system that the workers have.
I’m not treating unions as either a positive factor or a negative one here. They’re a fact of life in this industry.
OK.
The UK Southern experience really brings that home:
Driverless and completely unattended trains are the future, though not really for the reasons “management” might give. It is more about short headways, tight timetabling, safety and out-of-peak-hours service (the only situation where the bean-counters have a passable case about cost of staff).
In Paris, one of the reasons they’re automating is to avoid the constant strikes.
From far away, Southern looks almost like a conspiracy by Labour to show that privatized service is so bad that renationalization is the only option. Are they really that incompetent?
Not sure about that. First, there are very few strikes, and the usual ones are more a ritual: short, sharp and over very quickly (and rarely total, people can usually get to their destinations just slower; it’s all about sending a message to everyone that this system depends on all these ‘invisible’ workers). Second, while I am sure there were multiple reasons not least the overdue need to modernise Line 1, automation and the shorter headways was almost certainly a big factor. IIRC they have got it down to 85 seconds. Line 1 carries about 220m pax pa, the busiest of the Metro network (only RER-A has more with 300m pax pa.) and about 4x the busiest line in NYC (which run on multiple tracks).
Re Southern, it is simply the worst of the bunch. The entire privatized BR network gets heavy criticism (and is the most expensive system in Europe for pax). In France no one even thinks about the system because it just works without fuss and at reasonable cost. OTOH, Germany seems to be going down the British route and complaints are pretty fierce there too now.
Is there any fully-private passenger rail network that works well? (Let’s leave Tokyo & Japan out of this, because I find it contentious discussing private v public in their complicated ownership structures. Also I am not talking about where some small lines are private, like the Swiss system etc.) If private operation could work surely one would expect the NYC Subway or Amtrak to be fully private in the land of free-market capitalism? As I wrote earlier, the reason is that they are not permitted to, or it is technically and politically too difficult, to capture the huge value they create. Nevertheless they continue to run, even at a notional “loss” because the statement is so clearly true: they create immense value even if they themselves can’t capture it. The problem with Southern and the British privatized railways is that they merely capture government subsidies and so are not committed to service or proper partnership with their workers who run the system. As any number of railway experts have noted, the privatized system costs more (and in both government subsidy and fares, the most expensive in Europe) and performs worse than the old BR. And of course in the really critical part of the system (the track, maintenance and signalling etc) the private entity failed completely and had to be re-nationalised to avoid catastrophic collapse.
In terms of train drivers, we know from Vancouver Skylink that we really genuinely can automate the operation. The other classes of workers do remain necessary. If you want to pick a fight profitably, getting rid of the drivers seems like it does make sense. If you don’t want to pick a fight with them, offer them a different job. If they’re unwilling to take that offer, maybe it is worth picking that fight.
I use Paris and not Vancouver as my main go-to example here, because SkyTrain has been automated from the start, whereas the Paris Metro is automating lines that ran manually for a century, which is more difficult.
On the other hand, most transit vehicle drivers drive surface transit (buses or maybe single-car trams) and not long trains. For example, in New York, where the subway:bus ridership ratio is more than 2, there are 3,600 subway train drivers and 12,000 bus drivers, as of a few years ago. Automated buses are in development, but so far the technology is not mature yet. Automated planes… well, that technology is mature enough for hobbyist photography and for bombing weddings in Yemen, but not yet for carrying passengers.
Commercial airliners can fly themselves. From what I’ve been told, to the gate.
One-person train operation would cut the labor costs dramatically if it were implemented on more metro systems. Right now only Chicago L, D.C. Metro, and Baltimore have OPTO systemwide in the U.S. The MBTA is close, having it on all 3 heavy rail lines. Their Green Line is supposedly up for consideration for OPTO when they get Proof-of-Payment implemented under the next-gen farecard upgrade in 5 years. But that’s it, outside of the people-movers.
On the T it takes 56 six-car trains on-duty to run the Red + Orange + Blue Lines at peak. Under OPTO, that’s a 1:1 staffer per train ratio: 56 employees. Pre-OPTO every 2-car married pair in the consist had a guard, meaning 3 people per train and a rush-hour staff of 168. Now, to avoid picking fights with the union most of the displaced guards got auto-promoted to inspector or other positions so in the short-term it made staff efficiency much worse. But because they cherry-picked the more senior staffers and aren’t replacing people who are (increasingly) leaving it’ll normalize back bloodlessly into a more efficient equilibrium over time. If/when they deploy it on the Green Line those 73 two-car trains with 146 required cars and staffers at-peak can have their workforce cut in half.
NYC Subway currently only does OPTO on the Franklin Ave. and Rockaway Park Shuttles, and on the far off-peak and weekends on the 5, G, M, and the A-Lefferts Blvd. Branch when they slash back to 4-car lengths. All other times and all other lines it’s a 2-person crew. Not much they can do about the bus ranks, but a 1000-person reduction in subway staffing is completely doable if they had the gumption to go for it. 1000 fewer subway guards making $60K and up is money the MTA can much better spend on a whole lot more bus drivers to sustain growth for that whole generation of time it’ll take for driverless tech to be applicable in a real-world setting. 1000 fewer subway guards is also a whole lot more maintenance staff to hire for catching up on their endless state-of-repair backlog. The efficiency gains don’t have to unilaterally go to cuts and lowering payroll. Proper apportionment of staffing resources is just as important, with any huge agency having plenty of gaps to fill.
Why 1,000? There are 3,000 conductors/guards on the subway in New York. Plus another 3,600 drivers, which would go down to about 2,800 if staff efficiency were as high as in London.
….this actually relates to my comment about *lack of management* (where I used the example of my high school). MBTA has management. Chicago’s CTA has management. Both came up with deals to transition to OPTO and made it happen.
I see no evidence that the MTA in NYC has management. They seem to operate more on force of habit. This is even more apparent in their appalling refusal to comply with ADA regulations on renovations, which every other agency in the country complies with as a matter of course.
Speaking of ADA regs, on Twitter, John Robare notes that at North Quincy, a passenger in a wheelchair tried and failed to board unaided, and the train got delayed 10 minutes as a passenger stood in the doorway, trying to flag a conductor to activate the bridge plate.
That said, the example of Vancouver, where some of the rolling stock is 30 years old already, shows that big gaps that make it impossible for passengers in wheelchairs to board are not an inherent problem of subways. They’re just an inherent problem of poorly-run ones, e.g. in New York and Boston.
London Reconnections did a long work on Southern Railways which suggests most of it is not their direct fault, or at least understandable in a capitalist system.
http://www.londonreconnections.com/2016/railway-roulette/
I am not sure to which part of that almost 5,000 word analysis you are referring.
And does it matter? I mean privatisation was explicitly justified on the basis that private owners would be more capable of solving these complex issues–and at lower cost–than public authorities, which had apparently demonstrated over decades that they couldn’t. I put a lot of the UK problems of public service inadequacy (transport, healthcare, education, energy) to their culture post-Thatcher. When politicians remain hostile to public-run services (especially rail, a peculiar Thatcher aversion) and to funding public infrastructure (and yes it is exactly the same ethos that has led to absurd cost of such infrastructure in the Anglosphere). For example the “sickness issue” (of train drivers), as someone who has employed (research technicians) in both France and the UK, it is a real cultural problem in the UK. Even though Parisians are apparently among the world’s worst hypochondriacs, IMO it doesn’t affect work habits anything like the UK. (Of course it is also true that overall health of the UK is worse than France, eg. heart disease, obesity, diabetes etc). It is exacerbated by the “them versus us” attitudes, not to mention the mindless managerialism that infects everything today.
Another cultural issue peculiar to the transport systems of the UK, only indirectly alluded to in the article are “rolling stock” and “dwell time”: “When trains are overcrowded a train can be scheduled to have a one minute dwell time which can easily turn into three minutes in reality.” When I lived in Brighton a half lifetime ago the trains were still “slam door” which I found to be the most astoundingly inefficient arrangement imaginable in the modern world. I assume those are gone (though a dangerous assumption in the UK) and they are all auto-sliding doors. But then the London Underground have always been sliding doors and they too suffer the same dwell time problems: when you come to use it from any other city Metro it is extreme frustration waiting interminably at stations because, for “cultural” reasons, those doors are not firmly closed to tardy pax etc.
Here is the conclusion to the London Reconnections piece:
How then are the largely non-unionized and badly paid truck and bus drivers in the private sector in Germany to be explained?
Are other criteria valid for them?
…I’m honestly not sure. I know they’re unionized in France and the US, which are more or less the two lowest-union density first-world democracies.
Does Germany contract out bus operations? This can reduce unionization in the short run (just as the tech shuttles in San Francisco were initially non-unionized, before the unions successfully organized them).